At this point, a new top-tier AI model drops roughly every time I finish getting to know the last one. I've spent a slightly unreasonable amount of my maternity leave trying them all.

For most people, though, the problem isn't access to smart models. It's figuring out what to do with each new one before the next one arrives.

The default is to open the new model and give it the same work you gave the old model. Write this email. Summarize this document. Help me brainstorm.

Maybe it does the work a little better. But that doesn't really tell you what changed.

After listening to this episode of The AI Daily Brief and finding Daniel Miessler's list of prompts to run whenever a major new model drops, I realized I have a bit of a ritual for this.

Here's what I recommend doing every time a smarter model arrives.

1. Ask what just became possible for you

Don't start by handing the model a task. Start by asking it where it can raise the ceiling on your work.

This works best if it can access some real context about you: ChatGPT's memory, a connected Google Drive, your desktop files, a project folder, a GitHub repository, or whatever contains the actual mess of your work.

If it doesn't have that, give it a five-minute voice dump. A ramble with lots of context is more useful than a beautifully written but underspecified prompt.

Then try:

"Look across everything you know or can access about my work. Based on the capabilities of this new model, rank the five tasks or projects where it could create a meaningfully better result, or make something possible that wasn't before. Use evidence from my actual work, explain why each is newly viable, and give me a ready-to-run first test."

The difference between asking "What can you do?" and "What can you now do for me?" is enormous.

2. Give it a question big enough to expose its intelligence

Daniel Miessler's prompt list includes questions so large that I would previously have felt faintly ridiculous asking an AI to answer them.

That's rather the point.

A few adapted versions worth trying:

• Self-model audit: Review everything you believe about me. Where is your model of my identity, goals, voice or preferences stale, aspirational or inconsistent with my recent behavior? Propose the exact corrections.
• Big picture: Look across my projects, strengths, interests and the changes happening in my field. What should I focus on if I want work that is both fulfilling and lucrative? Give me a concrete recommendation, not a menu of vague possibilities.
• Where am I most wrong?: Take my biggest current bet and build the strongest possible case that I'm wrong about it. Which assumptions am I treating as facts, and what evidence should change my mind?

I've already run the first two with Sol. The self-model audit caught that my AI was still treating old business plans and an old maternity-leave timeline as current. That stale context was quietly shaping every recommendation it gave me.

A smarter model is a good excuse to clean up what your AI thinks it knows about you, then use that improved context to ask a much bigger question.

3. Run your personal litmus test

Every serious AI tester seems to have one oddly specific task they give every new model.

Ethan Mollick has his otter on an airplane using WiFi. He has run versions of that prompt for years, so you can watch AI image generation progress from melted piles of fur to photorealistic otters operating laptops.

Claire Vo has built a more formal four-part benchmark that tests every new model on writing a PRD, building prototypes, debugging code and sounding human.

Mine is an app I've wanted for years that would make it incredibly easy to capture the little memories of my kids and eventually turn them into something beautiful. Models have been able to build functional apps for a while. None has yet solved the trust, continuity and behavior-change parts well enough that I would actually keep using it.

Your test doesn't need to be scientific. It just needs to be:

• Something you care about
• Difficult enough that previous models haven't nailed it
• Consistent enough that you can notice real progress

Run it every time. Once all the models can do it easily, make the test harder.

4. Reopen something that never quite worked

One of the best ways to see what has changed is to return to an old project that a previous model almost got right.

Not a total disaster. An idea that was promising, but kept falling short because the execution, product judgment or technical lift wasn't quite there.

This week I rebuilt an old app of mine called Board of Advisors with Sol. In conversation, it figured out that this shouldn't really be a conventional standalone app at all. It should be an MCP server, essentially a connector that lets people use it directly through ChatGPT.

It then built most of the connector system and a much better-looking interface around it. I haven't fully tested the finished result yet, so the jury is still out. But it understood the product I had in my head and made an architectural leap that previous versions hadn't.

Try:

"Here is an old project or attempt that never quite worked. Here is what disappointed me about it. Reconsider the underlying problem from first principles. Is the old approach still the right one? What would you build now that wasn't practical or possible with the models available then?"

Don't only ask the new model to improve the old answer. Give it permission to question the entire approach.

5. Give it a bar, boundaries and permission to keep going

Smarter models can work longer and take more initiative. That is wonderful right up until they confidently change the wrong thing or take an action you wanted to approve.

OpenAI's current prompting guidance recommends giving the model the context it needs, the few constraints that genuinely matter and a clear point where it must stop for approval.

Nathaniel's episode also covers a newer pattern: giving the model a measurable bar for success, then letting it work in a loop. It creates something, checks it against the bar, finds the biggest gap, fixes it and checks again.

A reusable version is:

"Here is the outcome I want: [outcome].

We are done when: [specific, observable success criteria].

Keep these things unchanged: [important constraints].

Do not send, publish, purchase or take any external action without my approval.

After your first pass, test the result against the success criteria. Identify the biggest remaining gap, fix it and repeat until the bar is met or you are genuinely blocked."

"Make it high quality" isn't a bar.

"A new user can complete the core action without any explanation" is a bar.

"Research this thoroughly" isn't a bar.

"Every material claim is linked to a source, and conflicting evidence is surfaced" is a bar.

The clearer the finish line, the more useful the model's newfound tenacity becomes.

The most important part: raise your ambition

The easiest thing to do with a smarter model is give it your existing to-do list.

That can still save time. But it misses the bigger opportunity.

The most interesting work may not currently be on your list at all, because you've assumed it would be too hard, too technical, too expensive or too time-consuming for you to attempt.

So finish your new-model ritual with this:

"Look across my responsibilities, ambitions and ideas. What work am I not attempting because I assume it is beyond my skills, time or resources? Which of those assumptions might no longer be true because of this new model? Pick the most valuable possibility and help me design the smallest real test I could run this week."

The real test of a smarter model isn't whether it writes a better email.

It's whether it changes what you believe you're capable of taking on.

I'm on maternity leave. This newsletter was drafted by the AI agents I set up before going offline. Read about my AI-powered maternity leave experiment here.

One of my favorite things about AI is that it's a mirror. Give it context you already have, your emails, your Slack messages, your call transcripts, and it reflects back things about you that you'd never spot on your own.

One of the best places to point that mirror is your career.

Before Watch Me AI, I ran a career-design company called Crew, built around a program we called the Visionary. It was a program for people trying to figure out what they wanted from their career and what their next move should be. I loved watching it work: people sat with a coach and a small group, heard others reflect back what they'd said, and slowly got to "this is what I actually want."

And I watched the same problem sink people over and over. Career design is work. Being that self-reflective, step after step, takes time, patience, and usually money, and most people don't have all three. They'd start strong and stall out long before the insight arrived.

AI collapses most of that work. Not the thinking, the digging: the signal-finding, the synthesis, the writing-up. It's not a replacement for a coach or for sitting quietly with the hard questions. It's a shortcut, and a genuinely good one.

Here are three mirrors to try this week. Notice what they have in common: zero homework. You already made all of this context. You're just finally reading it.

1. Point it at your meetings

If you use Granola or any meeting note-taker, connect it to your AI (or export a few months of transcripts) and ask:

"Based on my meetings, what are my top three strengths? What do people keep coming to me for? Where do I sound most alive?"

You've been generating this evidence all year. Nobody's ever read it back to you.

2. Point it at your calendar and sent mail

Ask:

"Looking at my last 90 days, what work do I keep volunteering for, and what do I keep avoiding or pushing off?"

The gap between those two lists is your career compass, and it's sitting in tools you open every day.

3. Point it at the job market

In my old program, people would finally get clear on what mattered to them and then ask me: "Okay, now which jobs actually match my list?" Back then the answer was weeks of manual research. Now: paste your must-haves and two or three real job postings, and ask:

"Score each role against my list, and tell me what I'd need to probe in an interview to be sure."

A scorecard instead of a gut feeling.

That third question is where I'm taking this next.

I'm rebuilding that program as Career Design with AI. Seven steps, self-paced. At each one you feed the mirror context you already have (your meetings, your praise, your calendar, your resume), the AI does the synthesis, and you confirm or correct what it reflects back.

You finish with a one-page Career Canvas: what matters to you, your real strengths, and two or three named career directions with a next move for each. Then you put it to work, scoring real openings against your Canvas instead of your mood. Career clarity and an actionable path, in a couple of focused evenings, not the months of homework that used to stall everyone.

I'm opening a small founding round before it ships. $39, about half the planned price, fully refundable any time before it ships, no questions. Founding members get it first and shape the first version.

Reserve a founding spot for $39

And if a pre-sale isn't your thing, run the three mirrors anyway. That part's free, and it works.

This issue was drafted by my AI agents and approved by me from my phone. The experiment journal has the full story.

With all the AI talk, I realized I never actually updated you all on the baby. He's here!

Louis Henry Mueller, born June 2nd, and everyone is doing well. The hospital even offered to install the latest Anthropic chip directly into his brain to save him the whole interface era. Phew.

Which is a funny way into what I actually want to write about, because the rest of the AI world is waiting on the next big thing too.

Every morning I open the news to see if Fable 5 is back. It isn't.

Quick context if you missed it. Fable 5 is the new Anthropic model that came out a few weeks ago, and was available for about two days. Then the U.S. government issued an export-control order, and Anthropic had to switch it off for everyone, worldwide. It's been dark since.

OpenAI's newest model and Google's are in the same holding pattern, either delayed or handed to a tiny set of approved companies while the rest of us wait our turn. So the most powerful AI in the world right now exists, it works, and almost none of us can touch it.

I won't go into the geopolitics of these decisions, this post is about living in this weird, in-between moment.

Why I'm this attached to one model

Fable 5 was the first AI that felt like a real strategic partner instead of an agreeable junior staffer. It pushed back on me. I'd float an idea for the business, and instead of agreeing, it would tell me where I was wrong, which is the thing I've always wanted from these tools. Paired with the latest Claude Code updates, it could run on its own for way longer and do better work than anything before it.

My personal litmus test is a little app I've been building to save all the funny things my kids say and do, and Fable got me the best version of that yet. At the same time, it was giving me the sharpest guidance on my business I've gotten from any AI. It was also clearly going to supercharge this maternity mode/autonomous business experiment.

The walkway stopped

For two years, the deal has been that AI gets better roughly every week. A new model, a new tool, a new thing you can suddenly do. You get used to the ground moving under you. It's kind of like standing on one of those moving walkways at the airport. You're still walking, but something underneath is carrying you faster than your own legs ever could.

Right now the walkway has stopped. And when a walkway stops, you feel it in your body, this little lurch, because you were counting on the boost. That's where I am.

The current models are still good. But we had a taste of the next step, and now everything feels a half-beat slower than what we know is possible, and it's almost not worth starting anything big until the better thing comes back.

What are you going to do on your AI break?

Nathaniel Whittemore, who does the AI Daily Brief, has been covering this whole saga, including the news this week that Fable might be back online within days. He framed the moment as a question: what are you going to do on your AI break?

My answer?

Not much. I'm on leave, on the couch, watching Netflix with a baby on me, and for the first time in ages I feel zero FOMO about the AI news. It's great timing.

And while I'd normally encourage y'all to do as much with AI as you can, for this holiday week, I'd say feel free to get outside, enjoy the long weekend/world cup/Wimbledon, and wait for the new models to completely change the world. Again.

Hi Reader,

When people hear "advanced AI skills," they picture a certificate. A course you finish, a badge, some technical thing you either have or you don't. So a lot of people decide they are behind before they have even started.

The numbers say the opposite of what that fear assumes.

PwC just put out their 2025 Global AI Jobs Barometer, and they looked at close to a billion job ads to do it. Workers with AI skills earned a 56% wage premium last year. The year before, it was 25%. So the premium more than doubled in twelve months. In the same report, productivity in AI-exposed industries grew about four times faster than everywhere else.

Read that again, because the word "skills" is doing a lot of quiet work in that sentence. It sounds like it means coding, or prompt engineering, or a tool you have to master. It mostly doesn't.

I am not technical. I have said that in basically every newsletter, and I keep saying it because it is the whole point. The skills that earn the premium are not technical either. They are behavioral. They are about how you work with the thing, not what you know about it.

The clearest way I can describe it: the people getting that raise treat AI like a teammate, not a search box. Everything else follows from that one shift. Here is what it looks like in practice.

Treat it like a teammate, not a search box

A search box gives you one answer to one question and the conversation is over. A good teammate you go back and forth with. You tell them what you are actually trying to do, they push back, you correct them, they get sharper. That loop is the skill. Most people stop after the first reply and decide AI is fine but not amazing. The premium lives in the second, third, and fourth turn.

Give it context

A new teammate is only as good as what they know about you. Same here. Hand the AI your real documents, your goals, your past work, the way you actually talk. I have a whole running context that my agents pull from, and it is the reason this newsletter sounds like me and not like generic AI. The slow work of feeding it context is most of what makes it work.

Make it easy to hand things off

I do not type to my AI nearly as much as I talk to it. Getting your thinking out of your head and into the tool quickly, with as little friction as possible, is its own small skill, and it compounds. The easier the handoff feels, the more you actually use it, and the people earning that premium are the ones using it constantly, not occasionally.

Save what works

When a prompt or a workflow lands, keep it. Do not rebuild it from scratch next week. Reusable beats clever. A handful of saved workflows you trust will outperform a fresh burst of inspiration every time.

Go deep on one tool

I used to chase every new tool that launched. It felt productive and it was mostly a distraction. Going deep on one tool you know well beats hopping to the shiny new one every month. Consistency is the quiet skill that turns into the premium.

None of those five require you to write a line of code. They require you to change how you work, not what you know. That is good news, because behavior is something you can start today.

There is a real version of this happening already. Fortune ran a piece in May about solo founders using AI to do the work of entire teams, one person running what used to take a dozen. That is the 56% premium showing up in the wild. It is not going to people with the fanciest credentials. It is going to people who built the habit.

If you only do one thing, pick the first skill: next time you open AI, do not stop at the first answer. Go one more round.

In this newsletter, I'm sharing my setup. The agents that are working autonomously while I'm out. There are six of them so far:

The Newsletter Agent

This is the one writing to you from next week on. Every Sunday night it reads through my idea log, scans AI news and the people I follow, checks my old newsletters so it doesn't repeat me, and brainstorms topics for the new week. It then picks its favorite topic, drafts the whole thing, and sends me an email which I can approve from my phone.

The part that took the longest was the voice, because I trained it on how I actually write and talk, my old blog posts and the way I dictate into my phone, and not some polished AI version of me. It's got a post on how I built this lined up for later.

The LinkedIn Agent

Takes whatever the newsletter is about that week and reshapes it into a few LinkedIn posts, the same ideas in different forms, so I keep showing up there while I'm offline. It also creates net new posts based on the news and other topics I typically cover.

AI Mollie

My AI twin. She lives on my site and has access to some of my work and content so she can answer questions while I'm out. She is the lite version of the AI Mollie bot that lives in my Business Playbook and has my FULL library and AI brain behind it.

The Revenue Optimizer

This one owns the money side, which as a solopreneur on leave with nobody paying me is kind of the whole point. It keeps the things I've already built working, the lead generation tools, the products, the traffic to them, and it watches the numbers and proposes what to do next. For example, which offer a given week should feature, how to word a call to action, what to test. It hands all of that to me as ready-to-approve recommendations, and nothing goes live or gets priced or reaches the list without me signing off. It is also the newest of the bunch, so it is the one I'll be watching the closest.

The Research Agent

Reads the AI world so I don't have to keep up with it while I'm out. Every week it goes through the news, the new tool launches, and what the people I follow are saying, and it pulls out the handful of things actually worth writing about. It is how the newsletter stays current and timely instead of just recycling what I already think, and the Newsletter Agent starts from whatever it finds.

The Metrics Agent

Keeps an eye on all the numbers I would normally be refreshing myself, the subscribers, the opens and clicks, the sales, how the LinkedIn posts did, the traffic to my products, and every Friday it turns all of that into a short plain-English email that tells me what happened, what moved, and what to try next. I don't have to go digging through a dashboard, I just get the read, and it hands the same data to the Revenue Optimizer so that one can act on it.

What I approve, and what I never automated

Nothing publishes without me seeing it first. The agents draft, I sign off. And some things I deliberately left off the table for the whole leave: no client work, no live meetings, no changing prices or ad spend, no agent replying as me in private. The drafting is automated. The judgment isn't.

This experiment launched on Tuesday, so I'm giving it several weeks of autonomy to see how it's doing. I plan to check in later this month to see what needs human optimizing. I will keep y'all posted.

If all goes according to plan, by the time you're reading this, our second child will have arrived earlier this morning. As I write this, I could not be more excited to meet our little guy.

Here's the second big thing: I'm not pausing my business while I'm on maternity leave.

Instead, I've spent the last few weeks building a system that will keep this newsletter going, keep LinkedIn updated, keep generating income, and keep me learning from what happens, all while I'm offline with a newborn.

I'm calling it Watch Me AI: Maternity Leave Mode.

It's the biggest experiment I've run.

Why I'm doing this

A few reasons:

• I love running this business. If everything just stopped for three months, I'd have serious FOMO.
• I also like earning money. As a solopreneur, I don't have a company paying me during maternity leave. I want to see if I can keep revenue going while I'm away.
• I love AI and automations. I've been excited to try this exact experiment for months. Having a baby is the forcing function.
• I hope this inspires other moms, parents, and caregivers. If a working parent sees this and thinks "I could do that too," it was worth doing.

What it looks like

What's running on auto: The Newsletter Agent and LinkedIn Agent handle drafting. An AI Mollie bot handles questions, points people to resources, and keeps the conversation going. I've also added monetization and marketing layers to my existing products and content, so the things I've already built can keep working harder while I'm out.

A few things happening at once.

• The newsletter still ships every Tuesday morning. LinkedIn posts Monday through Thursday. The topics stay the kind of thing you've come to expect from Watch Me AI, all generated by AI Agents.
• An AI Mollie bot handles questions, points people to resources, and keeps the conversation going.
• I've also added monetization and marketing layers to my existing products and content, so the things I've already built can keep working harder while I'm out.
• Agents run periodically to check my earnings and optimize across my products for revenue.

What's actually off: my consulting work. No live meetings, no deliverables, no deadlines for three months. That part is genuinely paused.

The whole system has learning loops built in. The idea is that it can self-regulate, surface what's working, and maybe even grow on its own over the three months.

I'm not planning to be 100% removed. Once we find a rhythm with the baby, I envision using Wispr Flow to tell Claude what to do from my phone, probably while nursing or on walks. The goal is for most things to run on their own. The whispering part is for the fun stuff I want to bake in on top.

I also hope to go deeper on my AI craft and AI-mom projects. The first one just launched: a Family Crest Generator. More on the rest later.

I'm sharing the full journey here: molliemueller.com/maternity-mode.

What I'm excited about

Three things:

• Seeing what's actually possible. Most people would say you can't keep a one-person business publishing, selling, and learning while the founder is offline for three months. I want to find out.
• Learning a ton along the way. Building this has already taught me more about agent design and content systems than any single project I've done. Operating it for three months will be its own education.
• Inspiring someone, hopefully. If even one parent sees this and decides to try something they would have written off, that's a meaningful outcome.

What I'm worried about

I had way less time to set this up than I hoped. I was up late Monday night orchestrating agents instead of sleeping. So this is a real experiment in every sense: I'm not sure I did enough setup before today, and we're about to find out.

I'm worried about building in public and failing. Telling everyone I'm running this experiment makes it harder to quietly pull the plug if it stops working. I'd rather be honest than perfect, but the public part of "public experiment" has its own weight.

I'm worried about losing interest. Three months is a long time. The thing I've been most excited about for the last few weeks might not be the thing I'm most excited about by August. That's how most projects die.

I'm worried about it becoming a distraction from my family. The whole point is to be there for the new baby and my daughter. If the system keeps pulling me back to the laptop, it's worse than just pausing the business.

And I'm worried about the judgment that working moms get for all the things working moms get judged for. Whether I'm doing too much. Whether I'm doing too little. Whether wanting to keep working at all means I'm not committed enough to my kids. Whether automating the business means I'm replaceable. The whole catalogue.

I'm telling you these because if I kept them to myself the newsletter would feel less honest. And the honesty is most of the point.

What this could mean

Plenty of founder parents I know have either ground themselves into exhaustion during leave or stepped back from work that matters to them. There's been no real third option.

If keeping a business breathing while you're not at the wheel is possible, it's not just parents and founders who benefit. Anyone who needs a long step back, for health, family, sabbatical, grief, any reason at all, could use that playbook.

I'm building this in public so the lessons are visible. I really hope it works.

What's next

This morning, my second baby arrives.

Later this week, the next newsletter goes deep on how everything works behind the scenes: what the agents actually are, what's automated versus human-approved, what's not happening at all.

After that, every newsletter will be generated by my Newsletter Agent, reviewed and approved by me before it ships.

If you know another parent with parental leave on the horizon, a friend headed into sabbatical, a colleague burned out and dreaming of a break, or anyone who's wondered whether stepping back is actually survivable, forward this to them.

I hope this playbook gets used by many.

I was in The Wall Street Journal this week in a piece about people whispering to their computers. This is funny, mildly embarrassing, and completely accurate.

I use Wispr Flow constantly. I dictate prompts, emails, messy first drafts, strategy notes, and instructions to coding agents. I use dog walks for verbalizing brainstorms, and car rides for lengthy to-do lists. And yes, you can sometimes find my husband and me both muttering to our laptops from the couch.

Apparently I am not alone. I saw a LinkedIn thread this week where someone was making the case that keyboards may soon feel like CD-ROMs: useful for their time, but obviously not the endpoint. That sounds dramatic, but I think there's something real there. The bigger story is not Wispr Flow. The bigger story is that the keyboard is starting to lose its monopoly as the main interface for knowledge work.

Typing is too tidy

The chat box asks us to package our thoughts before AI ever sees them. It rewards the polished question, the clean prompt, the already-organized idea. But that is not how most thinking starts.

Thinking usually starts messier than that: "Wait, I think what I actually mean is..." or "This might be dumb, but..." or "The thing I'm frustrated by is..." or "I can't quite explain it, but something feels off..." That messy middle is often where the signal is.

Voice captures more of it. When I dictate, I give AI more context because talking is lower-friction than typing. I include the aside, the caveat, the emotion, the contradiction, the thing I would have edited out if I had to write it cleanly. And AI is often better when it gets the messy version first. And certainly faster at turning that into structure than I am.

From dictation to captured context

Wispr Flow is one version of this shift: solo thoughts becoming text. Granola is another version: conversations becoming context.

I'm fascinated by what you can do once your meeting history becomes searchable and promptable. The feature I've been playing with most is recipes, which are basically reusable prompts that run on top of your transcripts.

A few favorites recently:

• Coach Me Matt gives leadership coaching based on the Mochary Method. I've been using this to stay focused on my zone of genius and notice when I'm drifting into work I should probably delegate or avoid.
• Create LinkedIn Post turns a meeting transcript into a few possible LinkedIn posts. Not every conversation needs to become content, but some calls contain a useful idea I would otherwise forget.
• Find the Signal looks across a broader set of meetings and tries to identify a pattern no single meeting would reveal. This recently helped me notice a client concern I had moved past too quickly.

AI can now reason across my conversations, and it does so with zero typing required. This feels like a significant part of the shift away from keyboards.

The interface is moving closer to the thought

For a long time, the workflow was: think, type, organize, ask, receive output.

Now it is becoming: talk, capture, process, act.

The bottleneck in AI use is often not model intelligence. It is input friction: the gap between what is in your head and what the machine can understand.

The more that gap shrinks, the more useful AI becomes. A blank chat box is useful. A chat box with your dictated thoughts is better. A chat box with your meeting history, decisions, objections, patterns, and unresolved questions is much better.

That is when AI starts to feel less like a vending machine for answers and more like a thinking layer across your work.

Where this gets weird

Arjun Arora posted this week about Neuralink-style brain-computer interfaces. His point was that the most disruptive interface may not be another app or device, but a direct line between our brains and computers. In other words, the "input layer" could eventually move from keyboard, to voice, to thought itself.

Most people find the brain-chip version of the future insanely alarming. I get that. There are real questions about privacy, consent, control, and what it means to let technology get that close to thought.

But if I'm being honest, my first reaction is more excitement than fear.

Because if you follow this trend to its logical extreme, the endpoint is obvious. We keep trying to reduce the distance between thought and output. The keyboard was one interface. Voice is another. Meeting capture is another. Brain-computer interfaces are the more provocative version of the same question:

What happens when the machine can get closer to the thought before we have translated it into words?

That is exciting. It is also unsettling. And it makes the current moment feel like a transition point.

Today, I am whispering into my laptop. Tomorrow, maybe the interface is closer to thought itself.

For now, I am mostly grateful for anything that helps me more easily get ideas onto paper (or into this newsletter).

How do you feel about the new voice paradigm? And where do you land on the scale of alarmed to excited about brain chips? I'd love to hear.

Codex is OpenAI's coding agent and very similar to Claude Code. You give it access to a folder of files, tell it what you want to build or fix, and it works through the codebase for you.

This is not a tool most people need.

That is actually the point of this newsletter.

Whenever a new frontier model launches, the internet immediately becomes very dramatic. Everyone wants to know: Is this the one? Do I need to switch? Am I falling behind? Is Claude over? Is ChatGPT back? Do I need to spend my weekend migrating my entire AI life?

My honest answer: probably not.

Most people can carry on with their lives.

But if you are doing hard, technical, multi-step work, especially building digital products, GPT-5.5 is worth paying attention to.

Most people do not need GPT-5.5

Let me say this carefully, because I do not mean it in a patronizing way.

For most everyday AI use cases, you probably do not need the most powerful model available.

If you are using AI to draft emails, summarize meetings, brainstorm content ideas, plan a trip, organize your week, write a LinkedIn post, come up with dinner ideas, or think through a business problem, the top models are already very good.

Claude Opus is good. ChatGPT's prior models are good. Gemini is good.

At this point, for a lot of everyday work, the difference between the best models is less important than how much context you've given them, how well you know how to work with them, and whether you actually use them consistently.

This is not because GPT-5.5 is unimpressive.

It is because the baseline has gotten so high.

You do not need a Formula 1 car to drive to Trader Joe's.

And most people are driving to Trader Joe's.

Where I actually feel the difference

That said, I am not only using AI for everyday tasks.

I am a non-technical person who somehow keeps finding herself building digital products.

Client tools. Internal systems. App ideas, etc.

And in that world, I can feel the difference.

This is where GPT-5.5 through Codex has been impressive.

Not in a "wow, that answer was 7% better" kind of way.

More like: the work is moving faster, I am intervening less, and it is catching more issues before I have to point them out.

That is the kind of model improvement that matters.

What I'm liking about it

Here's what feels different so far.

It has better momentum.

Codex is faster than Claude Code. This is partly due to speed (it's significantly faster on the same tasks) and partly due to autonomy.

Claude Code is very powerful, but it constantly pauses to ask for permission. I understand why. It is a good safety behavior. But in practice, it means I have to babysit.

Codex gets through more work before handing it back to me. That makes it feel much more useful for the way I actually work: give it a task, go do something else, come back to progress.

I often tell it "I'm going to sleep now, keep working" and it will work for 2-3 hours while I sleep without getting stuck. I've heard other peoples' go for 8+ hours.

The context compaction is a big deal.

With Claude Code or Opus, I can see when I'm approaching the context limit, and this is very stressful. It means I'll have to ask it to write a handoff brief, open a new chat, paste the brief, and then spend time reorienting it.

I always lose something in that process.

Codex compacts the conversation and keeps going. For long product builds, that removes the "oh no, this chat reaching its context limit" stress and improves quality.

It catches more issues.

This is the biggest quality difference I'm noticing. On a complicated client product, it is catching more bugs, flow issues, and "if we change this, that other thing breaks" problems before I have to point them out.

That is where model quality becomes visible.

Not in one perfect answer, but in fewer corrections over time.

Why is it better than Claude Code?

I've heard that Sam Altman's strategy of investing heavily in data centers (i.e. capacity) is paying off relative to competitors. Apparently Anthropic just doesn't have the compute power that OpenAI has, which means its products can't keep up.

I have not fact checked this. And I wouldn't be surprised if the next Claude model leapfrogs 5.5. The pace of all these companies is insane.

So, should you switch back to ChatGPT?

Here is my current answer: probably not.

If you are using Claude and you love it for writing, thinking, planning, strategy, or general business work, you do not need to abandon it.

For those use cases, model choice matters less than context, habit, and using the tool consistently.

But if you are using AI to build, debug, operate across files, clean up messy outputs, or carry a complicated project forward over hours or days, GPT-5.5 in Codex is worth testing.

That is where I'm feeling it.

Not in one perfect answer. In the accumulation of fewer interruptions, better continuity, faster progress, and less babysitting.

Sharing a few recent projects in hopes they inspire you on what you can create with AI. For yourself, your family, as gifts, or just to push your creative confidence.

The watercolor house portrait

The first project was a gift for a friend's birthday.

I found a photo of her house on Redfin and used Gemini to turn it into a watercolor painting.

It was surprisingly fast. I think I made a couple of tweaks, but not many. It came out really beautifully, and personal and giftable in a way that felt much more special than buying something random.

I printed it on nice paper, framed it, and gave it to her.

My prompt for this (in Gemini Pro Nano Banana) was something like: "do a watercolor version of this house" (with image attached).

Our house, but make it an architectural sketch

After the watercolor gift worked so well, I started trying a version for our house.

For ours, I'm trying for more of an architectural sketch style, something more classic and line-drawn.

This has been harder. I have not nailed it yet, What I love about doing this at home is that the stakes are low. I can experiment. I can print a test. I can decide it's not right. I can try again later.

It doesn't have to be precious.

The neighborhood postcard pack

The next project was a little postcard pack of neighborhood landmarks.

At first I thought I could convert existing photos into watercolors, same as my friend's house. However I didn't have any good photos of the landmarks I wanted. So instead, I asked Gemini to create watercolor paintings of those landmarks based on a few bad Google search images. It did brilliantly.

Now I have four pieces that feel like a cute little neighborhood postcard pack. I printed one of each and framed them for a gallery wall in our house.

I'm also thinking about printing sets and giving them as gifts to friends in the neighborhood.

My daughter's name art

The project I'm working on now is art for my daughter's room.

You know those illustrated name signs people make for kids? Or when you're on vacation and there's someone on the boardwalk who will paint the letters of your name in a cool style?

That's the idea.

I already had four matching frames in her room with art in them. I've been ready to swap that art out, so now I'm making one illustrated letter for each letter of her name.

I've been experimenting with different styles, and the one I'm currently into is over-the-top floral sculptures. I found inspiration images on Pinterest (mainly of wedding floral designs) and shared those with Gemini with a prompt about turning them into "the letter A".

What I love is that this can evolve.

If I decide in six months that I want a different style, I don't have to buy all new art. I can just make another version with ChatGPT, print it on 8.5 x 11 paper, and swap it into the frames.

That is such a small thing, but it makes decorating feel more flexible.

You can try something. Live with it. Change it later.

For this project, I used the new image gen in ChatGPT and it was awesome.

Why this feels different

I've written before about using AI for images, holiday cards, coloring sheets, storybooks, and visual assets. But this feels like its own category.

The difference is the physicality. These are not just digital outputs. They become gifts, wall art, keepsakes, postcards, framed prints.

And the combination of AI plus a good printer is weirdly powerful.

AI gives you the image.

The printer makes it real.

The frame makes it feel intentional.

Suddenly, you are not just generating pictures. You are making objects for your life.

A few ideas to try

If you want to play with this, here are a few easy places to start:

• Turn a friend's house into a watercolor portrait
• Make a postcard set of your neighborhood
• Create illustrated letters for a kid's room
• Turn a favorite vacation photo into a vintage travel poster
• Make a set of recipe cards in the style of an old family cookbook
• Create custom wrapping paper or gift tags
• Make a tiny gallery wall around one theme: your city, your pet, your favorite restaurants, your garden, your house

You do not need to become a professional artist. You just need to make one thing you like enough to put on a wall.

A note on printing

I have not owned a home-printer in a very, very long time.

Creating these AI images and seeing the possibilities for art, gifting, and product building got me to finally purchase a printer for our house. I went fancy since I wanted something that could print quality art prints and gifts (Epson 8550), and I have zero regrets. It's an incredible printer and you save money on ink. Paper, however, has been a significant additional expense since you gotta get the good stuff.

I could go on about my love of the printer, but you do not need a fancy printer to do these projects. You can get them printed at a local print shop or send them to Shutterfly or similar. Or even print on a low quality home/office printer just to get a feel for things before investing.

Note: the paper type makes a HUGE difference. So don't get discouraged if your art looks bad on regular paper. Happy to share more about my printer/paper learnings with those interested.

Have you made any AI arts & crafts? If so, I want to see them! Hit reply and send me a picture.

It's a new mode inside Claude built for visual work like slides, posters, marketing assets, app mock-ups, etc.

I've been in it for four days. Here's what I've been making, what's working, and what's not quite there yet.

How I started

Same way I start with every new tool: I asked Claude for ideas.

Prompt: "Claude Design just came out. Give me 10 personal ideas and 10 work ideas for things I could make."

It came back with a decent list and as I was just interested in exploring, I quickly picked a few.

Here's what I've tested/built so far

1. A magazine about our home life

One idea on the list was a fictional magazine about our family life. Claude called it the Mueller Monthly.

I copied Claude's brief into Claude Design and let it run. It took a while, way longer than creating visual assets in Claude Chat, but what came back impressed me. It was an 8-page magazine covering our remodel in a humorous, design-forward way. There was a feature on banquettes. A full-page ad for sophisticated drywall. A profile of our dog. And a spread about my new printer. All topics that I've discussed with Claude Chat recently.

I'm likely not moving forward with this idea because Claude doesn't have enough context on my home life to make the magazine content personalized and good. But, the design held up consistently across every page, the humor surprised me, and it did it all in one shot.

Takeaway: Claude Design is great at multi-asset projects with a consistent feel.

2. A series of neighborhood posters

Next I tried a poster series of favorite spots in neighborhoods we've lived in. I gave it a rough list, just neighborhood names and restaurant names. That's it.

It came back with 11 stylized posters. I can tell from the content on the posters it did some web research to figure out details like street name.

I don't love the style it started with, but I could easily change that and turn these into something I'd be happy to hang.

Takeaway: I can skip the reference-gathering step. Even on a low-effort brief, it makes something I'd actually want to print.

3. An app to capture the cute things my kid says

This is something I've been trying to solve for ages. Guessing others can relate.

My 2-year-old daughter does and says adorable things constantly, and I never capture them. I've tried voice notes, text notes, apps, pen and notebook, nothing sticks. It's a behavior problem more than a tool problem. I just can't get into a habit of capturing these moments.

I told Claude Design all of that. Instead of jumping into design, it interviewed me. What have you tried? Why did it fail? What might make you open the app?

Then it designed around my answers. It proposed a few different capture interfaces and let me see mockups for each. I'm currently leaning towards iMessage. It designed a nudge system. It mocked up a progress tracker showing a memory book filling up week by week, so the motivation compounds. It designed the physical year-end zine that would come out of it.

All of this was done in one shot and came out as something between wireframes and mockups. The "tweaks" feature makes it easy to try different styles and approaches, as does the main chat interface (of course).

Takeaway: Claude Design slows down before it builds. Lovable is fast, you give it a brief, it ships. Claude Design made me consider the UX and the behavior first. That's a real strength for anything where getting the design right matters more than shipping quickly.

4. Marketing assets for the AI Business Playbook

Then I turned it on my actual business.

I asked for marketing assets for the AI Business Playbook and gave it the public sales page URL and six screenshots from inside the product.

From those few assets it pulled my brand colors, my fonts, and my visual style. Then it built an Instagram post, a carousel, a LinkedIn graphic, a new email header, and a redesigned landing page hero.

Takeaway: Give it a little context and it extracts more than you'd expect. This is the fastest I've ever gone from "I have a product" to "I have a marketing kit."

A few things I don't love (yet)

No native image export. I wanted to download an asset as a PNG to post on LinkedIn, and the option wasn't there. It had PDF, HTML, PPT, and Canva, but no simple download as image. When I asked it how to download an image, it built me a custom export button which is overkill for basic functionality. I assume this gets fixed soon.

It's slow. Like Claude Code slow (not a bad thing). The upside is it does its own design reviews and iterates before calling something done. The downside is you can't really sit and watch. I've started treating it the way I treat Claude Code: send it a prompt, go do something else, come back when it's ready.

It devours credits. It operates separately from your regular Claude credits, and apparently we don't get many to start because I burned through mine in the first couple of days.

Who is this for?

Probably not designers. If you live in Figma and know what you're doing, this isn't your tool.

This is for the rest of us. People who need good-looking design, don't have the skill or budget to make it happen themselves, and want something better than a Canva template.

That's me. Probably you too.

Where does it fit in my stack?

I build almost everything in Lovable, end to end. Planning, building, iterating, shipping, hosting. Lovable does all of it.

Claude Design doesn't replace that for me. It can't host, and hosting is a big part of why I stay with Lovable.

But I can see using Claude Design for the initial stages of a project: mocking things up, exploring directions, getting the design right before I build. Then handing off to Lovable to ship. A lot of folks will hand off to Claude Code instead. Either works.

The bigger thing I'm noticing

Every new Claude product is another reason to give Claude more context about me.

When I played with the Mueller Monthly, the limit wasn't the tool. It was that Claude didn't know enough about my actual life to make it personal.

With my business, it's a different story. Claude already has my meeting transcripts (via Granola), my Google Drive, my website. So when I asked for marketing assets, it could do the whole brand lift without much direction.

The pattern is clear: the more context Claude has, the more it can do for me. And as it gets better at making things (designs, apps, strategies, campaigns), the payoff for feeding it context keeps going up.

Try it yourself

Claude Design is in your account if you have Claude Max. Open a new chat and try "design mode."

Start small. A poster. A one-pager. A brand moodboard. Something low-stakes where you can get a feel for it. Then work your way up to the bigger stuff.

I run this recipe after almost every meeting. My internal rule: if there's even a remote chance I said something smart on a call, I run the recipe. It takes about thirty seconds.

First: What is a recipe?

Granola is an AI meeting transcriber. A recipe is Granola's term for a saved command. Basically a prompt that's saved in my Granola account that I can use with one click or one "/recipe" command. Granola has a lot of cool ones listed in their recipe library (I think they call this a "recipe book" lol), and you can create (and share) your own.

What my LinkedIn recipe does

I use Granola to transcribe my meetings. After a call ends, I open the chat box on the meeting note, type a slash command, and pick a saved recipe called Create LinkedIn Post. Granola reads the transcript and gives me three full LinkedIn posts in my voice, each based on something I actually said.

The output is good enough that I rarely edit. Most days I copy one straight into LinkedIn. Some days I schedule all three.

The exact prompt

Here is the recipe. You can use this link (if you have Granola), or copy/paste the below text into your own recipe/prompt. You will want to swap the parts about my audience and voice for yours. That's the whole customization job, and it takes about two minutes.

You are my LinkedIn ghostwriter.

You'll be given the transcript (and any notes) from a single meeting.

Read the meeting and identify three distinct, high-impact ideas, stories, or frameworks that would make strong LinkedIn posts for my audience: leaders, solopreneurs, and busy professionals who want to use AI to save time, earn more, and work smarter.

For each idea, write a full LinkedIn post in my voice: Mollie Amkraut Mueller, founder of Watch Me AI.

Tone: clear, confident, practical, and a bit conversational; no emojis and no exclamation points.

Write in the first person.

Start with a strong hook in the first line.

Use short paragraphs and, when helpful, 2 to 4 bullets for clarity.

Make the insight concrete and tied back to the meeting content (what I noticed, what I learned, what I'm testing, etc.).

End with a simple invitation to reflect or engage (e.g., a question or "Curious how others are handling this."), without sounding salesy.

Do NOT reference the meeting itself, any private details, or that this came from a transcript.

Make each post standalone and ready to paste directly into LinkedIn.

Return your answer in this exact format:

Post 1: [full post]
Post 2: [full post]
Post 3: [full post]

How to make it yours

Replace two things and you're done:

• The audience line. Mine says "leaders, solopreneurs, and busy professionals who want to use AI to save time, earn more, and work smarter." Make it specific to who you actually want to reach.
• The voice instructions. Mine says no emojis, no exclamation points, conversational. Yours might be different. If you don't know your own voice rules yet, paste five of your favorite past posts into Claude and ask it to describe the patterns. Use that as your starting point.

Why this works

Most "AI for LinkedIn" tools generate posts from a topic prompt. The output is generic because the input is generic.

This recipe works because the input is a real conversation you just had. The ideas are already yours. Granola is just helping you notice which ones are worth saying out loud, and shaping them into a format that lands.

You finish a meeting with three posts you would have written anyway, if you'd had two more hours in your day.

Try it this week

Run it after one meeting this week. Tell me what comes out. Or better yet, tag me in your LinkedIn post. And if you have any favorite Granola recipes, please share. I'm making a list.

We started with the usuals: best AI tools for work, how to learn AI, etc., but what's really stuck with me from that conversation were the questions she saved till the end:

"Can I use AI to help with meal planning? What about laundry? Can it help me manage all those household things that eat up my entire weekend?"

She's a mom. She's busy. She's heard that AI can do all these incredible things. And she wanted to know if any of those things included the stuff she actually needed help with.

I want these things too. So I really wanted to have a better answer, but here's what I told her and still believe:

The honest answer

AI cannot do your laundry. It cannot unload the dishwasher. It cannot drive your kids to practice or fold a fitted sheet or remember that you're almost out of paper towels.

And despite what the headlines might suggest, it's not going to be able to do those things for a long time. The physical-world stuff, the stuff that requires a body and hands and being in a specific place at a specific time, is still way beyond where AI is today. The robots are coming, sure. But they're not coming to your kitchen this year.

I had this conversation months ago. My answer hasn't changed.

What AI is genuinely good at right now is work that happens on a screen. Writing, researching, building, planning, organizing information, creating things. It's a thinking tool, not a doing tool. And that distinction matters, because it changes what you should actually expect from it.

The reframe

Here's what I told my friend, and it's the same thing I'd tell you.

Don't try to use AI to replace the physical stuff. Use AI to make more money. Then use that money to pay a human who can actually do the physical stuff.

That might sound like a dodge. It's not. It's the most practical advice I can give.

AI is genuinely making it easier to earn income in ways that didn't exist two years ago. You can build a small software product without writing code. You can start a side business with less overhead than ever. You can automate the boring parts of freelance work and take on more clients. You can create content, build tools, launch services. The barrier to starting something has dropped dramatically.

The person folding your laundry doesn't need to be you. But the person building a new income stream probably does, at least at the start. And AI can help with that part right now.

So how do you find the idea?

Maybe it's turning your professional expertise into consulting or coaching. Maybe it's selling something you already make on Etsy. Maybe it's a small digital product, a template, a guide, a tool.

The options are out there. But from talking to a lot of people about this, I've come to realize that coming up with the idea is often the hardest part. Not because there aren't good ideas, but because it's hard to see your own skills clearly enough to know what's worth building around. And even when you have a hunch, making that idea feel real enough to actually start is its own challenge.

So, of course, I built an AI app to help with this.

It's called Spark. You give it your LinkedIn URL and answer a few questions on your interests and constraints (like how much time you have and what you're trying to accomplish), and it generates business ideas with real market context.

You could literally tell it: "I want to generate enough side hustle income to pay someone to do my family's laundry. I have about one hour free per week." And it will come back with ideas matched to your experience and constraints.

I'm not promising you'll be outsourcing laundry by next month. But it's a real place to start. It's currently totally free!

Try Spark here: idea-spark-3000.lovable.app

At some point I may need to charge for it, because it costs me money to keep it running. But for now, I just want to get as many people as possible trying out new ideas with AI, so I'm keeping it free for the short term.

Give it a go. There's no downside. And I'd love to hear if it sparks anything for you.

Who knows? Maybe by this time next year, you'll have enough side hustle income to outsource the laundry.

Today I want to tell you about my favorite one: the proposal generator.

The problem

This team runs a lot of new business meetings. They're good at it. They land a meeting, the partners have a great conversation, and afterward the team needs to turn that into a polished proposal deck to send the prospect.

Here's how that used to work:

An associate would gather all the notes from the meeting (their own, the partners', whatever context they had on the prospect) and then spend hours building a PowerPoint from scratch. They'd pull from past proposals and a bank of existing slides, but every deck still required significant manual assembly.

An external design firm was involved for the final polish, which added cost and time on top of that.

The thing is, most of these proposals are structurally similar to each other. There are standard slides that describe how the firm works, their track record, their team. Those don't need to be reinvented every time. But there are also sections that need to be highly tailored to the prospect: the situation overview, the communications strategy, the media targets, the pricing. Those parts need to be highly strategic.

But most of the associate's time wasn't going to the strategic thinking. It was going to the mechanics. Building slides, formatting decks, assembling the same pieces in a slightly different order. The kind of work that feels productive but isn't really where a smart person's time should go.

So we built a proposal generator in Claude.

What it does

The system lives inside a Claude Project. It has the firm's brand guidelines, a full catalog of their slide types, reference decks, and detailed instructions for how proposals should be structured.

Someone on the team pastes in their meeting notes and whatever they know about the prospect. Claude generates a written outline for the entire deck.

It maps out:

• which standard template slides to pull in and why
• where it needs to generate new custom sections based on this specific prospect's context

Then, Claude:

• writes the situation overview
• maps out a communications approach
• identifies relevant media targets
• suggests case studies to include
• and drafts pricing

That outline comes back in about two minutes. The team can review it, make changes, and adjust the strategy. As soon as they're happy with it, they hit go and Claude generates the actual PowerPoint presentation. Fully branded, properly structured, ready for review in three minutes.

The part that surprised me

I expected the time savings to be the headline. And they are. We're talking about a process that went from many hours to minutes. But what actually impressed me was the quality of the strategic sections.

The situation overview doesn't just summarize the meeting notes. It synthesizes them into an analytical narrative about the prospect's competitive landscape, their communications challenges, and why this firm is the right fit. When we tested it against a real prospect, it pulled in context about the sector, identified the right peer comparisons, and priced the engagement in line with what the team would have proposed themselves.

It's not ready to send to the client as-is. It still needs a human reviewing the strategy, adjusting the tone, adding judgment calls. But it gets the team to a strong first draft faster than they thought was possible. In their branding. In their voice. With their slide structure.

Why I'm sharing this

I go back and forth on how much client work to share in this newsletter. But I think this one is worth talking about because the pattern is so common.

Almost every company I work with has some version of this problem: a repeatable deliverable that eats up way more time than it should because no one has stopped to systematize it. Proposals, reports, client onboarding docs, pitch decks. The specifics change but the shape is the same.

The tools to fix this exist right now. Claude Projects, custom instructions, brand guidelines as context. None of this requires code or a technical team. It requires someone who understands your workflows well enough to design the right system, and knows how to use Claude.

If that sounds like something your team needs, I'd love to hear about it. Hit reply and tell me what you're working on, even if it's just "we have this one process that's killing us." That's usually the right place to start.

It was part of SheBuilds, Lovable's initiative to get more women building with AI. For International Women's Day, they made the entire platform free worldwide. Over 120 community-hosted events across 40+ countries. Anthropic and Stripe partnered to give every participant free credits on top of it.

I've been involved with SheBuilds since their first season last fall, when I gave a talk on turning your builds into a business. So when IWD rolled around, I blocked the morning and started building.

By evening, I had two working apps.

Here's what I made:

The Women in AI Directory

This one had been brewing for weeks. In February, I posted a question on LinkedIn: "Who are the women in AI you're learning from?"

839+ women were nominated (and counting!). Builders, researchers, educators, executives, ethicists, policy leaders. The comment section turned into this massive, community-sourced list of people worth following.

But here's the thing about comment sections: they're hard to browse, impossible to search, and easy to lose. I didn't want all that energy to just disappear.

So I built a directory.

Every woman nominated in that thread now has a profile you can browse, search by name or role, and follow on LinkedIn in one click.

I built this using Lovable (what you see), plus a few Relay automations that helped me scrape the LinkedIn comments to get everyone's names, titles, nomination count, etc.

Building a women-in-AI directory on International Women's Day, using free credits from a company that goes above and beyond to support women was, needless to say, a bit meta. And felt very right.

And it's already growing on its own. People are sharing it, claiming profiles, adding their info. You build the container, and the community fills it.

Spark: the startup idea finder

This one came from a question I keep hearing: "I want to build something, but I don't know what."

I hear it from friends. I hear it in workshops. People are excited about AI-powered building tools, but they're stuck at step zero, the idea.

So I built a tool that takes you from "I have no idea" to a validated concept with an actionable playbook in about five minutes. I'm calling it Spark.

Here's how it works:

• Discover. You fill out a short form, your skills, what you wish existed, how much time you have, what kind of project you want (full-time business, side hustle, nonprofit, just for fun), plus your LinkedIn URL.
• Ideas. AI generates 8-10 personalized startup ideas based on your background, interests, and constraints. Each comes with a founder-fit score, problem statement, and target market.
• Builder's Brief. Pick your favorite and get a full playbook: step-by-step build guide, customer acquisition strategy, pricing, tech stack, estimated time to MVP, and a launch checklist.

Here's how I built it:

• Rambled about my idea into the Lovable chat box
• Built the Lovable pages in steps
• Copied an existing Relay.app automation and updated for this app's specific data needs
• Lots of testing

What these builds tell us about applicable AI

Both of these apps took me half-day to get to a version 1. And I am not a technical person.

A year ago, building a searchable directory with profiles, search, and follow buttons would have meant hiring a developer and frankly, I'd never have done it. It was cost-prohibitive and intimidating.

That's the shift. Building software is becoming as casual as writing a Google Doc. You describe what you want. You iterate. You ship.

And the most interesting part isn't the speed, it's what you choose to build when the barrier drops. I didn't build a productivity dashboard or a CRM. I built a community resource and a tool to help people get started. When building is free, you build what you care about.

If you've been sitting on an idea, a tool for your business, a resource for your community, a side project that's been living in your notes, the barriers are lower than you think. You don't need to code. You need an idea and a few hours.

If you've been reading this newsletter for a while, you know I've usually avoided telling people which general AI tool to use. My advice has always been: pick one of the leaders (ChatGPT, Claude, Gemini) and go deep. The memory you build up, the context it learns about you, the workflows you develop, that's where the real value is. Not the tool itself.

I'm changing that advice today.

It's time to switch to Claude.

I've been using Claude as my primary AI tool for two months now, after being a ChatGPT-first user for 24+ months. The gap has been consistent enough, for long enough, that I'm now comfortable recommending Claude directly.

Caveat: This is the AI world. ChatGPT could launch a new model tomorrow that changes the math. That has happened before and it will happen again. But two months is a long time in AI, and Claude has been meaningfully better for my work the entire stretch.

Why switch?

The short version: Claude is a better writer, it has stronger integrations with the tools I already use, and it can do full tasks for me, not just answer questions. The overall quality of what I get back is noticeably higher, and it has been for months.

But rather than just tell you why, let me show you what I'm actually doing with it, and how to easily make the switch.

What I'm actually doing with Claude

To give you a sense of what "switched to Claude" looks like in practice, here's how it shows up in my week right now.

Projects are my operating system.

I have a Project set up for every client, one for my online course, one for this newsletter, and one called "Business Building" where I do everything from financial planning to daily to-do lists. Each one has instructions and reference files uploaded, so Claude always has context.

Claude is my executive assistant.

Inside my Business Building project, I have one conversation I use many times a day. I've gotten Claude to build me a visual dashboard with tabs for my financial projections, weekly scorecard, pipeline, sprint plan, and to-do list. Every morning I open it and say "what am I supposed to do today?" or rattle off what's in my head, and we make a plan for the day together.

I build almost every document in Claude now.

Especially slides. Client updates, end-of-project summaries, live event decks, I create them all in Claude. It can export as Google Slides or PowerPoint, but my favorite method is HTML: I open it right in my browser and it works as a scrollable presentation. They look great, and the turnaround is minutes instead of hours.

Skills for instant branding.

Skills are like detailed, repeatable prompts saved inside Claude. I built one called "Watch Me AI Branding" with all of my brand rules. Now anytime I create a document in Claude, which I do constantly, I just say "apply my branding" and get a beautifully formatted version in one step. I've built these for clients too, and I recommend everyone create one for their own brand.

This newsletter was written with Claude Code.

Claude Code is the techie version of Claude, but is actually quite approachable. I have a workflow set up with all of my past newsletters, my voice guide, and a set of rules (has this topic been covered recently, am I repeating formats, that kind of thing). When I sit down to write, Claude already knows the full history and helps me brainstorm, draft, and edit in one place.

It connects to my other tools.

Claude integrates with Granola (my meeting note-taker), my email, my calendar, and Google Drive. I can say "pull together the full story of my project with this client" and it goes into Granola, finds every relevant meeting, and pieces together the whole narrative. No copy-pasting. No switching tabs.

None of this requires technical skills.

How to switch (it takes about 10 minutes)

The biggest barrier people mention is memory. All the context ChatGPT has learned about you over months of conversations. You don't want to start from zero.

Anthropic just solved that.

• Step 1: Open Claude's memory import. Go to your Claude settings (Settings > Capabilities > Memory) and click "Start import." Or look for the "Import memory" card on your Claude home screen and click "Get started."
• Step 2: Export your ChatGPT memory. Claude gives you a prompt to copy. Paste that prompt into ChatGPT, and it will list out everything it knows about you, your preferences, your work context, your recurring tasks, in one block of text.
• Step 3: Paste it into Claude. Copy that output, paste it into Claude's import box, and click "Add to memory." That's it. Your context comes with you. No file exports, no technical steps. Copy, paste, done.

A few notes: this feature is still experimental, so Claude may not pick up every detail. It also focuses on work-related context, so very personal memories may not carry over. But for most people, it covers the important stuff.

A few more ways to ramp up:

• Set up a Project. Create a Project in Claude and upload the documents you reference most, brand guidelines, your offer details, writing samples, whatever you use regularly. This gives Claude persistent context that goes way beyond what memory alone can do.
• Use it for real work for one week. Make yourself use Claude as your default for five straight days. Try creating presentations, build a brand skill, ask it what else it can do for you. I'm positive you'll see the difference.

It's the most honest conversation I've had publicly about how I work. Including the part where I admitted that when I started consulting, I had no idea what to charge clients so I just asked Claude to benchmark it for me.

We also got into:

• The $1,500 diagnostic that replaced a dev shop spinning its wheels for 5 months
• How I turn one meeting transcript into 10 different assets after almost every call
• How I built an automation that scrapes 700 conference websites weekly for a PR agency
• My real feelings about that MIT study claiming 95% of AI projects fail

Listen on Spotify

Some of you may have seen my LinkedIn post last week where I asked people to name women doing interesting work in AI. The response was beyond anything I expected -- over a thousand comments and counting. Hundreds of names.

I'm compiling all of them into one resource and I'll be sharing the full list soon. But I know there are more names that haven't been mentioned yet.

If there's a woman working in AI who deserves to be on this list -- including yourself -- go add her name in the comments.

Link to post

The more complete we can make this, the more useful it becomes for everyone.

What I'd do if I were job hunting right now

A friend of mine is job hunting. He emailed me with a few specific questions, and I sent him a completely overkill 10-page response (AI x career design = my happy place). My process? I opened Claude, started brain-dumping everything I'd want him to know using Wispr Flow, and 15 minutes later I had the full, structured strategy document: AI Powered Job Hunting in 2026.

It was long. Way more than anyone asked for. But when I looked back at it, three moves stood out as the ones that actually change outcomes and they all use AI in ways most job seekers haven't considered yet.

Whether you're looking yourself or know someone who is, here's where I'd start.

1: Build content around the job you want (not just the work you've done)

The insight: Most job seekers either post nothing on LinkedIn or post generic "open to work" content. The smart move is to reverse-engineer a content strategy from the actual roles you're targeting.

The tactic: Take a job spec you're excited about. Feed it into Claude along with your past work -- old decks, reports, presentations, emails you're proud of. Ask it to generate content ideas that position you as someone already working in that space. Then have it draft 3-4 LinkedIn posts from your existing material, angled toward that role.

The payoff: When a hiring manager checks your profile, they see someone already living in the work -- not just applying to it. And when you follow up after an interview ("I actually wrote about this recently"), it's a very strong signal.

Core point: Your content strategy should be pulled toward where you want to go, not just reflecting where you've been. AI makes this easy because it can bridge the gap between what you've done and what you want to do.

2. Ask for specific intros, not vague favors

The Insight: Most people ask their network some version of "do you know anyone who could help me?" That's too vague. It puts the work on the other person and usually goes nowhere.

The Tactic: Here's what works better. I call it the Friendlies Method.

Make a list of 5-10 people who would happily do you a favor -- close friends, former colleagues, well-connected people in your orbit. For each one, go to their LinkedIn profile and browse their connections. Make a list of anyone they're connected to who would be valuable for you to meet.

Then text your friend: "I put together a list of people I'd love to connect with -- do you know any of these folks?"

The payoff: That's an easy yes. You've done the research. They just have to say "sure, I know Sarah, let me connect you."

The AI part: use Claude to research each person you want to meet. Have it write a short, personalized reason for the intro -- something specific about their work that genuinely interests you. If you're sending more than a few, you can batch the whole thing: paste your list into Claude, give it context about your background and goals, and have it generate a customized intro request for each pairing.

I wrote a full guide on the "forwardable email" format that makes these intros land well -- "The right way to ask for introductions". The short version: write the email so your connector can hit forward, add one line, and send. Make it effortless for them.

Core point: Do the research yourself, then hand your connector something easy to act on. AI handles the research and personalization at scale.

3. Record every conversation and mine it

The insight: Every networking call is an asset -- if you capture it. Most people have a great conversation, send a generic "great to meet you" follow-up, and forget 80% of what was said within a week.

The tactic: Use an AI meeting recorder. I use Granola. After each call, paste the transcript into Claude and ask it to do three things:

• Draft a follow-up email that references specific things you actually discussed. A personalized note that mentions the topics you covered is far more memorable than "great to chat."
• Pull out names and leads. Ask Claude to scan the transcript for every person mentioned, every company referenced, and anyone the other person suggested you talk to. These are warm leads you'd otherwise forget by tomorrow.
• Flag content ideas. You probably said smart things in that conversation. Ask Claude to identify anything you said that could become a LinkedIn post or short article. After a handful of calls, you can write a summary piece: "I've talked to 10 people in [your industry] this month. Here are the 5 things everyone is thinking about." That kind of post invites engagement from exactly the people you're trying to meet and signals that you're already in the room.

Core point: Conversations are raw material. AI turns them into follow-ups, leads, and content -- all from one transcript.

The bigger picture

None of this is new strategy. Networking, building visibility, getting warm intros -- that's always been how people land good jobs. What's different now is the execution speed.

The research that used to take an evening takes minutes. The follow-up emails that used to feel like homework write themselves. The content you never had time to create is suddenly possible because AI can bridge the gap between what you know and what's on the page.

The strategy hasn't changed. The excuse for not doing it has.

That's the gap the AI Business Playbook was built to close.

What it is

The AI Business Playbook is a self-paced, modular library of AI workflows, automations, and playbooks designed for business operators, not developers. It's the exact system I use to run my own business, packaged so you can copy, adapt, and deploy it in yours.

It's not a collection of generic prompts. It's not a course where you watch someone explain what AI is. It's a working system you can install into your business in pieces, starting with whatever is most useful to you right now.

What's inside

The playbook is organized into three layers:

The Tech Stack

Deep dives into the specific tools that power modern business automation. Each module walks you through setup, use cases, and real workflows:

• ChatGPT and Claude — Advanced prompting, use cases, and how to get genuinely useful output (not the generic stuff most people settle for)
• Granola — Using AI meeting intelligence for pre-meeting prep, coaching insights, and automated follow-ups
• Relay.app — Building invisible automations that handle scraping, inbox filtering, and multi-step workflows
• Lovable — Spinning up client pages, roadmaps, and internal tools without a dev team
• Canva, Napkin AI, and Gamma — Turning text into polished visuals instantly

Functional Playbooks

Plug-and-play systems organized by business function. You pick the area that matters most and install the workflow:

• Marketing and Content — The "1 Idea to 10 Assets" engine. Turn a single voice note into newsletters, LinkedIn posts, and video scripts.
• Sales and Growth — Modern prospecting without the hard sell. Automate RFI responses, lead enrichment, and personalized outreach.
• Client Delivery — Proposals in minutes instead of hours. Turn meeting transcripts into instant deliverables like decks, diagrams, and summaries.
• Operations — The "Morning CEO Brief." Automate SOP creation and build systems that give you time back.

The Executive Layer

This is the part most AI resources skip entirely. AI isn't just for tasks, it's for thinking:

• Strategic Planning — Revenue forecasting, quarterly planning, and prioritization with AI as a thinking partner
• Executive Coaching — Automated weekly reflections, pattern detection in your leadership style, and accountability systems
• Ethics and Control — A practical guide on privacy, data protection, and knowing when not to use AI
• Implementation — A 30-day plan to build your personalized stack and get your team on board

Who it's for

The playbook is designed for three types of people:

The Scalable Solopreneur. You run a one-person business, consultant, creator, coach, and you've hit a ceiling. You don't want a big team. You want AI to act as your marketing manager, executive assistant, and sales analyst so you can focus on delivery.

The Boutique Agency Founder. You sell time and expertise. You need to protect your margins by getting client work done faster. You want to cut delivery time in half and impress clients with speed, without sacrificing quality.

The Full-Stack Leader. You might work inside a larger company, but you operate like a founder. You own a P&L or lead a function. You're drowning in meetings and admin. You want AI to buy back your time so you can focus on strategy and leadership.

Who it's not for

This isn't for people looking for "get rich quick" hacks or passive income schemes. It's not a PDF of 10,000 generic prompts. And it's not a theory course about the history of large language models. The playbook is about opening the tools and building real automations.

About the creator

I'm Mollie Mueller, founder of Watch Me AI. I'm a former IDEO Strategist and a founder who has built and sold AI-powered products. I've helped over 1,000 professionals figure out how to use AI in their work, and I've coached leaders at Fortune 100 companies on leveraging AI for strategic advantage.

Most people teaching AI are tech enthusiasts. My background is in human-centered design. I don't look at AI as code, I look at it as a tool to design a better way to work. That's the perspective the playbook is built on.

Pricing and access

The AI Business Playbook is currently $497, one-time payment, with lifetime access to all current and future modules. The system is actively updated as tools evolve and new workflows are added.

There's a 14-day satisfaction guarantee. If it's not what you need, email me for a full refund.

Get the playbook here: https://molliemueller.com/courses/ai-business-playbook

She knew AI could help her business. She'd heard the pitches, seen the demos. But she was stuck at the same question I hear from almost everyone: where do I actually start?

A friend had told her to go hire an engineering team on Upwork. Build something custom. But she had a hunch that was overkill. Expensive, and premature. Why commit that kind of money before you even know what to build?

What she actually wanted wasn't an engineer. It was clarity. As she put it: "I didn't know how to evaluate the work. What I actually wanted to know was what I should build."

So she hired me to figure that out with her.

She was already sitting on a goldmine.

This coach had been recording her coaching sessions (with client permission, of course) using an AI meeting note-taker. Every session from the past few months was fully transcribed and saved. She knew there was more she could do with all of it. She just needed help figuring out what to build and how.

If you take one thing from this newsletter, let it be this: if you're recording your meetings, you have raw material for almost everything you need at work. Follow-ups, content, analysis, client support, career development, team management. And if you're not recording yet, start. (I use Granola. Link here.)

That library of transcripts became the foundation for everything we built together.

What we built

We built three things. All from meeting recordings, simple automations, and some careful prompt work. Total cost was a fraction of what an engineering team would have charged.

1. A coaching feedback analyzer

This one was about trust and support. She wanted a way to evaluate her own coaching. Things like her talk-to-listen ratio, the types of questions she was asking, whether she was advising too much versus coaching. The kind of feedback that's hard to get when you work alone.

We started by building a prompt that analyzes her session transcripts against her own coaching framework. She could run it after any session she wanted to review. One click, instant analysis.

Once she trusted the output, we automated it. Now she gets a short coaching report every week. She doesn't have to read every one. But it's always there, running quietly in the background, giving her the kind of honest feedback most leaders never get.

2. Automated session summaries for clients

This was the biggest client experience upgrade. After every coaching session, an automation drafts a warm summary email with the commitments the client made, key insights from the conversation, and a reminder that goes out a week later.

The emails land in her inbox as drafts so she can review before sending. Her clients now get structured follow-up and accountability without her doing any additional work. Before this, follow-ups were inconsistent or nonexistent. Now they're automatic.

If you're in any kind of client-facing work (coaching, consulting, advisory), think about what it would mean if every client got a thoughtful follow-up within hours of your conversation, every single time.

3. A client-facing AI assistant

This was the most ambitious piece. We built a web app where her clients can chat with an AI trained on her coaching methodology and their specific session history. Each client has their own login, their own data. No cross-contamination, which is critical when you're working with sensitive executive conversations.

Now her clients get coaching-style support between sessions. And she has a full admin dashboard to manage everything herself.

The whole project took about six weeks and cost a fraction of what she'd been quoted by engineering teams.

How we approached it

The tools weren't the hard part. Two things mattered much more:

Strategy before building. We started by mapping her business: where her time went, what her clients needed, where AI could actually make a difference versus where it would be a distraction. The first sessions were almost entirely conversation. That's the part most people skip, and it's the part that determines whether anything you build actually sticks.

Quality before automation. Every tool was tested manually first. She refined prompts, reviewed outputs, lived with the results for a while. Only after she trusted the quality did we automate. This matters. If you automate bad output, you just get bad output faster.

Where she is now

She has three AI workflows she owns and can modify without technical help. Her clients are getting a level of follow-up and between-session support that didn't exist before. She has a feedback system that gives her the kind of self-evaluation most coaches only get from a supervisor.

And other coaches in her network have started asking how she did it.

If this sounds like you

Most people I talk to are in the same place this coach was. They know AI could help. They've tried a few things. But they haven't built anything real into their business because they're stuck on the "what" and the "how."

Three ways to get started:

• If you're recording your meetings (or willing to start), you already have what you need. Look at your last few transcripts and ask: what am I doing with this? Pick one workflow. Maybe it's better follow-ups. Maybe it's prep for your next conversation. Maybe it's pulling out patterns you're not seeing in real time. Build a prompt for it. Test it a few times. See if the output is useful. That's exactly how we started with this project, and it's how I start with every client.
• The AI Business Playbook walks you through the tools and workflows I use with clients like this, step by step. If you want to go deeper on your own, start here.
• The AI Workflow Diagnostic is a 90-minute session where you walk me through your operations and I bring the questions, the research, and a clear-eyed assessment of what's possible. You walk away with a recommendation doc: specific workflows, specific tools, realistic timeline. Whether you implement with me, someone else, or on your own. Book a free 15-minute intro call here.

It's a little embarrassing to admit this out loud. We tend not to talk about money, at least not this directly. But I want all of you to try asking your AI this question, because I have found it so useful and so empowering.

Let me start with a recent example.

I asked Claude: "Given everything you know about my business and goals, could I be earning more money?"

Claude has access to my meeting transcripts, my business goals, my website, my newsletter metrics, and my bank account. So when I ask this question, it gives me a specific, personalized response.

(Note: Most of this takes five minutes to connect. Start by asking Claude: "How do I connect [this tool/source] to Claude?")

Here's what it flagged this time:

• I was underpricing my digital product. Based on my credentials and audience engagement, it argued I should be charging significantly more and laid out the reasoning.
• I was undervaluing sponsored content. It pointed to my newsletter's open and click rates (which are well above industry average) and told me my floor should be much higher than what I'd been quoting.
• I wasn't stacking recurring revenue. It mapped out a progression: newsletter → digital product → group workshop → high-ticket consulting. Each layer feeding the next.

A different week, the same question caught something I'd missed. It explained I'm decent at getting attention, new subscribers, LinkedIn views, strong open rates, but my conversion from reader to customer was relatively low. I don't have a traffic problem, I have a funnel problem. It listed out specific things I could do to fix it.

That kind of feedback, super personalized, with the depth of a really good business advisor, would be worth it on its own. But it gets way better, because now Claude can actually do things on my behalf.

I'll look at the recommendations it gives for how I can make more money, choose a few, and say "do this for me." And off it goes.

For example, when it flagged my conversion problem, the conversation went something like this:

Me: Analyze all of the data, tell me my current conversion numbers, and tell me what I should do to fix it.

Claude: [comes back with a list of ideas]

Me: Which of these can you do for me?

Claude: [paraphrasing here] Basically all of them. I'd recommend starting with a new welcome sequence. Here's what it could look like.

Me: Great, please do it.

It wrote five emails for a welcome sequence. Once I reviewed the copy and made a few tweaks, I told it to get this set up in Kit, my newsletter tool. It opened up a browser, went into Kit, set up the emails, told me when it was ready for my review. I hit publish. Done.

(Two weeks ago, this kind of action required Claude Cowork. Now Claude Chat is actually doing a lot of this on its own.)

Here's what's changed in the past year that's made this so powerful.

A year ago, you could certainly ask AI "how can I make more money?" and you would have gotten back decent advice. But it wouldn't have been as personalized to you, because it wouldn't have had as much context on you and the work you do.

Over the past year, our AI tools have gotten way better at knowing about us. They connect to more of our tools, they remember past conversations, and they talk to each other. It's a lot less manual work to feed it context and a lot more "it magically already has it."

And on the action side, you saw it above, Claude doesn't just give you ideas anymore. It implements them.

This continues to improve. Let me give you an example literally from last week:

Granola now connects directly to Claude

Granola, which you know is my favorite tool for meeting transcriptions, just came out with an MCP integration, which in human-speak means Granola now connects to other tools like Claude.

So now, when I ask Claude "how can I make more money," it's not only looking at my business planning threads and my bank account and my website. It can comb through my sales calls and understand how my pricing conversations are going, without me needing to manually input any of that.

If you don't use Granola, the bigger point is this: AI tools are starting to talk to each other, and that trend is accelerating.

A note on tools: I'm now using Claude for about 90% of my AI work, which is a big shift from two months ago when it was 90% ChatGPT. This is the first time I've moved so dramatically from one to the other. It's very possible ChatGPT launches something that rebalances the race, but for now, my wagon is hitched to Claude. Sharing because y'all are always asking about my go-to tools.

Not a business owner?

If you're reading this and thinking "great, Mollie, but how does this apply to me?", I've got ideas for you. Try asking Claude or ChatGPT:

• "What's my strongest case for a raise or promotion?"
• "Where am I spending time at work that isn't advancing my career?"
• "If I wanted to increase my income by 20% this year, through my current role, a side project, or both, what would you recommend given what you know about me?"
• "What's a low time-commitment side hustle that could earn me $2k/month?"

Then say: "Great, how much of this can you do for me?"

The key is the same: with enough context about your situation and goals, AI can help you make moves (big or small) you might not otherwise have made.

It's a tool that turns your LinkedIn profile into a fantasy career map. Every role becomes a realm. Every pivot becomes an adventure.

You paste in your LinkedIn URL and instantly get a personalized, illustrated map of your professional journey.

I've been tinkering with LinkedIn profile digests and visual ideas like this for a while, but this is the first time it's come together so easily.

Here's what's wild: I didn't write any code. I didn't make any of the art. I just connected four tools and let them do the work.

The stack:

• Lovable built the website
• Kit captures the form submissions
• Relay.app orchestrates the automation
• Nano Banana (Gemini's image generation) creates the maps

When someone submits their LinkedIn URL, Relay reads their profile, writes a prompt, sends it to Nano Banana, gets the image back, and emails it to them. Automatically. I do nothing.

None of these tools require technical skills. If you can write a prompt and click some buttons, you can use them.

Why this matters beyond a fun side project:

A year ago, this wasn't possible. The image generation wasn't good enough. The tools didn't talk to each other smoothly. You'd need a developer to stitch it together.

Now? Four tools crossed a threshold at the same time. And suddenly someone like me can build something like this in a weekend.

I think we're going to see a lot more of this, people turning passion projects into micro-businesses because the barrier to building just dropped dramatically.

I'm already imagining where this could go: anniversary maps that trace a couple's journey together, graduation maps for students, wedding maps showing how two paths converged. Maybe partnering with a print shop to turn these into framed gifts.

That's all future stuff. For now, I just wanted to share what's possible.

Want to see your career as a quest map?

Try it here: careerquestmaps.lovable.app

It's free while I'm experimenting. If you love yours, post it on LinkedIn and tag me. I'd love to see your journey.

And if you get your map and think "my team should do this together," send them the link. Comparing career maps might be the most fun team-building exercise I've accidentally invented.

But most of us aren't developers. So I've been watching from the sidelines, wondering when the rest of us would get something like that.

Two weeks ago, we did.

Anthropic launched Claude Cowork and I've been deep in it ever since. Today I want to give you the honest rundown: what it is, what it's actually good for, and whether it's worth paying for.

What Cowork actually is

Regular Claude Chat is a conversation. You ask, it answers. You're still doing the work.

Cowork is different. You give Claude a task, and it goes and does it.

It can browse the web. It can open apps. It can read your Google Drive. It can take screenshots, click buttons, fill out forms. It can chain multiple steps together without you babysitting each one.

It's less like a chatbot and more like an employee.

AI that can browse the web and take actions has existed for a while. But it required specialized software and a lot of patience. This is the first time it's been this easy, this powerful, and inside a tool you're probably already using. That's what makes this a big deal. Anyone can do this now.

Let me make this tangible

I know that sounds abstract. So let me show you what this actually looks like with a few real examples from the past week.

1. Auditing my free guides

I have a folder in Google Drive full of free guides I've created over the past year. Things like "Getting Started with Lovable" and "How to Use AI for Meeting Follow-ups."

The problem is, AI tools change fast. Some of those guides are probably outdated but going through each one, researching what's changed, and deciding what needs updating has been sitting on my to-do list for months.

So I handed it to Cowork.

I gave it access to my Google Drive folder and said: review each guide, research whether the product has changed, and tell me which ones need updates and what specifically should change.

It opened each guide. It went to the product websites to check for new features. It came back with a prioritized list: high priority updates, low priority, no changes needed.

Then I told it to implement the changes. And it did. It went back into each guide and made the updates itself.

That would have taken me an entire afternoon. Cowork did it while I was on a call.

2. Reviewing and updating my website

I built my personal website in Lovable. It's been a few months since I really looked at it, and I wanted to make sure it was optimized for my current goals: selling the Playbook and booking consulting clients.

So I told Cowork: go to my website, review every page, and give me recommendations for what to improve.

It went through the site page by page, took screenshots, and came back with a detailed analysis.

Then I told it to open Lovable and actually make the changes.

It did. It opened Lovable, typed in the prompts to update my homepage copy, and hit send. I just watched.

3. Updating a massive conference list

A client came to me with a spreadsheet of 500+ conferences. Each row had a URL, and the dates and locations needed to be updated for this year.

Doing this manually would be brutal. Open URL, find the 2026 dates, copy them over, repeat 500 times. That's hours of mind-numbing work.

So I gave it to Cowork. Go to each URL, find the updated dates and location, and fill in the spreadsheet.

It just started going. Opening each site, finding the information, updating the row, moving to the next one. It even added a "changes" column so we could review what it did, and color-coded rows based on the type of change and its confidence level. Most were marked high confidence, and when we spot-checked, it was highly accurate.

This is the kind of task that used to require a VA or an intern. Now it's a prompt. (And yes, this is exactly the kind of thing I help clients figure out.)

A few tips if you try it

Use Opus 4.5. It's the newest model and noticeably better than the others. You might burn through credits faster, but the quality difference is worth it.

Create a local folder. Cowork works best when it has a folder on your desktop it can save files to. I just made one called "Claude" and point it there.

Be specific about your goals. The more context you give (what you're trying to achieve, what success looks like, any constraints) the better the output. Treat it like you're briefing a new employee.

So when do you use Cowork vs. regular Chat?

The line is blurry. And getting blurrier.

Something I've noticed in the past week or so is that Cowork-style capabilities are showing up in regular Claude Chat. Like web browsing, file creation, and multi-step reasoning. Just this morning, Chat used web browsing to add new tags in Kit so I can better segment my subscribers.

My prediction: Anthropic is going to merge Chat, Cowork, and Code into one interface eventually (which in AI time likely means next week).

My rule right now: start in Chat and see how far you can get.

Graduate to Cowork when you need Claude to:

• Browse multiple websites and pull information
• Access files in your Google Drive or desktop
• Do multi-step projects where each step depends on the last
• Take action inside other apps

Which means even if you're not paying for Max, you should start learning how to work this way. Giving Claude tasks instead of questions. Letting it browse and research. Thinking of it as an employee rather than a tool.

This is where AI is heading. The sooner you get comfortable with it, the better.

Is it worth $100/month?

For me, yes. Easily.

The time I've saved on the projects above, plus a dozen others, has been worth way more than $100. And it's enabling me to do things I simply wouldn't have done otherwise, like that guides audit.

That said, it's still buggy. I've had days where it barely worked. It's improving fast (noticeably better than even a week ago), but it's not a polished product yet.

If you're curious but not sure, try it for one month and see how much you use it. If you're not getting value, cancel.

And if $100 feels like too much right now, keep using Chat. A lot of this capability is coming there too.

Which means I've been deep in "finishing mode": recording videos, editing, tweaking pages, all of it.

But here's the thing about building an AI Playbook with AI: it's so fun to build things that sometimes you stop building THE thing and start building side projects that help you build the thing.

Case in point: instead of recording last week, I spent an hour building this:

It's a progress tracker for all the playbook sections. Each row is a section, each colored button is a deliverable/step. Click a button, it fills in. The progress bar moves. Dopamine hits.

I now check this thing constantly and it's genuinely helping me ship faster.

How I built it

Two or three prompts in Claude Chat. That's it. Not Claude Code, not Cowork, just regular Claude.

My starting prompt:

"I need a simple visual checklist so I can track which sections I've recorded, built, and what's open. Here are all the sections. I want to be able to check them off as I complete them."

Claude built it. I clicked "preview" right there in the chat, and started using it immediately. When I want changes, I just ask.

I didn't write any code. I just had a conversation.

Watch the 3-minute tutorial

A peek at what's inside the Playbook

You can see the section names right there in the tracker:

• Tool Playbooks: ChatGPT, Granola, Lovable, Relay, Wispr Flow, and more
• Functional Playbooks: Content Creation, Marketing, Sales, Operations, Strategy, Revenue Planning, Biz Coaching
• Plus, Ask Mollie: An AI bot trained on everything I know. Ask it questions as you go

Each section is a video walkthrough plus a cheat sheet with prompts you can steal.

Today, I'm sharing the audio side: the tools that let me talk instead of type.

This shift has been huge for me. I now do most of my work by speaking. Emails, AI prompts, meeting follow-ups, even this newsletter, I'm talking my way through all of it.

Here's what's in my audio stack:

Wispr Flow (my favorite new tool)

I'll be honest: I was skeptical of paying for a dictation tool. I was already using voice memos on my phone and occasionally ChatGPT's audio mode. Why would I need another thing?

Then I tried Wispr Flow.

How it works:

You download it as an app on your computer. Hold down the Fn key, start talking, and when you release the key, Wispr Flow transcribes what you said and pastes it directly into whatever app or browser you're in. Perfect transcription. No copying and pasting required.

My old workflow wasn't bad (record into my phone, copy the transcript, paste it into ChatGPT). It wasn't hard or time-consuming. But Wispr Flow is so effortless that I was instantly hooked. Within two days, I'd blown through all my free credits and upgraded to paid. I haven't looked back.

What I use it for:

• Talking to Claude, ChatGPT, Lovable, Relay, basically any AI tool with a chat interface
• Writing emails, Slack messages, documents
• Drafting newsletters (I'm using it right now to get these thoughts on the page)
• Anywhere I used to type, I now speak

Wispr Flow also has a mobile app that works as an alternative keyboard, so you can get the same quality dictation on your phone. I find that using Wispr Flow and submitting the text to ChatGPT gives me much better results than using ChatGPT's built-in voice feature. Plus, I find those voices kind of annoying.

This is one of those tools where I've heard from multiple people: once they start using it, it's a whole new world, and they go all in fast.

Try it here (free month with my link)

Granola (my meeting note-taker)

Granola is an AI meeting assistant that joins your calls and creates transcripts and summaries automatically. You've probably heard me rave about it before, it's still my go-to for every meeting.

Why I love it:

It saves me hours every week. I use my meeting transcripts to create presentations, write follow-up emails in minutes, build project plans, and prep for future meetings. One meeting can turn into ten different deliverables.

(All of those workflows are in the AI Business Playbook, on sale now.)

Why I chose Granola over other note-takers:

The UX is clean and simple, I actually enjoy using it. And at this point, I've been on it for a year, so all my meeting history lives there. I'm locked in.

Granola (free credits with this link)

Voice Memos (the simple one)

I still use Apple's native Voice Memos app regularly.

When I use it:

When I'm out on a walk and want to capture a stream of ideas. I open Voice Memos, ramble for 10-15 minutes, then later drop that transcript into ChatGPT to organize and structure it.

I could technically do this in Wispr Flow's mobile app, but for long-form idea capture, I find it easier to just hit record and go.

Sometimes the simplest tool is the right one.

Cool Tech Worth Knowing About

These aren't core to my daily workflow, but they're impressive and worth experimenting with:

NotebookLM Audio Overviews

Google's NotebookLM can turn your notes and documents into AI-generated podcast-style conversations. I've heard students love this for learning complex concepts. I don't have a strong use case for it personally, but the technology is genuinely impressive.

ElevenLabs (voice cloning & text-to-speech)

I've dabbled with this for creating voice content. Really cool technology that I could see myself using more in the future, especially for creators who want to generate audio at scale.

Transcript-based editing in Tella

Tella is where I record my videos, and it has a feature where you can edit video by editing the transcript, delete a sentence of text, and that section of video disappears. Super intuitive for cleaning up recordings.

Audio in Lovable projects

You can add voice instructions and audio features to anything you build in Lovable. I've experimented with this before and it's exciting that anyone can add this kind of functionality to a product without code.

Final Thoughts

The shift from typing to talking has changed how I work.

I move faster. I capture more ideas. And it just feels more natural, like I'm having a conversation instead of staring at a blank page.

Here's something I didn't expect: when I use my voice to give instructions to ChatGPT or Claude or Relay, it feels like I'm talking to an executive assistant. I no longer have a human EA (I use AI for most of those tasks now), but there's something about speaking out loud versus typing that reminds me of that relationship.

It's helped me reframe how I think about these tools. They're not just software. They're more like employees. And I think that mindset shift, treating AI as your team, not your tool, is the future of how we'll all work.

If you haven't tried voice-first workflows yet, Wispr Flow is where I'd start.

And if you want to see the full system, how I turn meeting transcripts into content, how I structure AI prompts, how I've built workflows that save me 5+ hours a week, that's all in the AI Business Playbook.

Get it here: The AI Business Playbook (40% off until next week)

I don't mean creativity in some abstract, art-school sense. I mean this:

Anyone with an idea can now execute on that idea in ways they never could before.

In work, building a company that would have required a team and outside capital two years ago. In life, tackling DIY projects you were scared to start, illustrating a children's book for your kid, designing a custom invitation instead of paying for a template.

We can now do so many things that were beyond us as individuals before AI.

2026 is the year the downstream effects of this unlock really start to show. Here's what I think we'll see.

1. Solopreneurship explodes

I'm a solopreneur now, largely because I can build an entire business on my own. Two years ago, I would have found that very challenging. When I built my last company, I needed a co-founder, an engineering team, and external capital, partly because I wanted to scale it big, but partly because there were so many things I simply couldn't do myself.

Now with AI, I can do all the things. Code a product. Create and execute an entire marketing strategy. Design beautiful ads. Time is my biggest constraint and increasingly, I'm building AI agents to solve that rather than hiring people.

It's never been easier to build a money-generating company by yourself. These tools both teach us how to do things and do a lot of that work for us. All the barriers have dropped.

Combine this with a white-collar job market that's already getting massively disrupted, and you get a wave of people going out on their own, hanging their own shingle, making their own money.

Here's a spicier take: I think there are interesting long-term political implications here. If we see a real shift toward solopreneurship, toward everyone needing to bring in their own bacon, that could lead to a more capitalistic mindset across the board. This is different from how today's white-collar workers think about work and economics. For many years, we've worked for big organizations that felt safe and secure, that provided benefits and stability. As a solopreneur, you give up that safety net.

But here's the thing: that safety net, that implicit promise, has already eroded. I recently read something from Greg Isenberg about how he's been spending time with Gen Zers, and they're so disenfranchised by the traditional work promise that many of them are starting out as entrepreneurs. They're more capitalistic by nature than us millennials ever were. The shift may already be underway.

2. Personal software becomes a real paradigm

People will increasingly build themselves small, sometimes disposable, mini apps rather than paying for third-party software.

I'm already doing this. When I needed to send out an invitation recently, my instinct wasn't to go to Paperless Post. It was to open Lovable and build a simple invitation with a tracker exactly the way I wanted it, for no additional cost beyond my existing subscription.

Last month I built a digital holiday card for my family instead of going to Minted and paying them for something similar. It was more fun, more custom, and more mine.

This will accelerate dramatically as more people realize how easy it is to build with AI. The mindset shift is: why pay for a generic solution when I can build exactly what I need in 20 minutes?

3. Personal agents go mainstream

We're in the middle of a big interface shift with AI agents.

Previously, building an agent meant stringing together different tasks in a workflow builder. It wasn't hard, but it was harder than just chatting with ChatGPT, and most people wouldn't do it.

Now, you build an agent the same way you talk to ChatGPT. You just say: "I need an agent that combs my email and preferred news sources every morning and sends me a podcast on what I need to know." And it builds it in front of your eyes. (If you haven't tried agent builders like Relay.app recently, go look, this is happening now.)

Given this change in ease, many more people will start building their own custom agents. We'll see much more specific, bespoke use cases emerge, and a much higher proliferation of agents in daily life and work.

4. We get comfortable collecting our own data

I've been writing about this for a while: we're entering an era where the return on personal data is extraordinarily high. The equation has completely changed, data capture is now effortless, AI surfaces patterns we'd never find ourselves, and crucially, it takes action on our behalf.

In 2026, I think we'll see a big shift in how comfortable people are with ambient data collection. Recording your meetings, letting wearables track your patterns, feeding your calendar and email into AI tools, all of this still feels a bit edgy to many people. But the career and life advantages are becoming too significant to ignore.

Case in point: Meta recently acquired Limitless, the company behind a pendant that listens and records your conversations. A year ago, this was 100% fringe, most people still shudder at the thought of wearing something that's always listening. But Meta buying it signals where this is headed. Combine that with the exploding popularity of Oura rings and other health wearables, and the trajectory is clear.

The health implications alone are massive. When your wearable can detect early signs of illness, optimize your sleep, and correlate your energy levels with your habits, the value proposition becomes undeniable. Expect this to tip quickly, within a few years, collecting this kind of personal data will be the norm, not the exception.

5. Digital-to-physical services emerge

Here's a gap I keep bumping into: I can now create a flyer, an invitation, a poster for my kid's room, a storybook, etc. fairly easily, and with a ton of fun in the process. But when it comes time to make that creation into a real physical thing I can hang on the wall or give as a gift? The solutions aren't great.

Current printing and fulfillment services weren't built for this new reality. They're not AI-native. They don't anticipate the volume and variety of custom creations that people are now producing.

Given this massive unlock in digital creativity, I think we'll see new services and companies emerge specifically to help us translate AI-created work into real-life artifacts. Someone's going to nail this, and it'll be a big deal.

6. Leaders get strict about AI adoption

"Use AI" won't mean "chat with ChatGPT sometimes" anymore.

I predict leaders will get much more adamant that their teams must:

• Implement entirely new AI-driven processes
• Adopt new AI tools beyond the basics
• Do their work in fundamentally different ways than before

If people aren't doing that, they'll get cut. We've already seen some leaders say this publicly. I think it ramps up significantly in 2026 both in public statements and, more importantly, in the internal expectations within organizations.

The bar is rising fast.

7. "Vibe coder" roles emerge inside companies

Here's one from Nathaniel Whittemore (host of the AI Daily Brief podcast): he predicts we'll see new roles emerge around AI-enabled coding, including something he calls "internal forward-deployed vibers," staffers meant to help different groups figure out how to use AI coding tools to improve what they do.

This resonates with me because I'm seeing massive demand for this already, just outside of companies. Organizations, individuals, everyone is hungry for someone who can come in and show them how to implement AI specifically for their function, role, or use case. Often because they don't have the awareness of what's possible, and also because they don't have the time.

It makes sense that we'd start to see this role formalized inside companies. The need is clearly there.

8. The time paradox

AI saves us time. I'm now much faster at many tasks, so I can get more done in a day.

You'd think this means we win back time for non-work pursuits. I've come to think otherwise.

Credit to Scott Belsky, who recently wrote about this in a way that crystallized something I'd been feeling. He invokes something called Jevons Paradox: when a precious resource becomes more efficient to use, we don't conserve it, we consume more of it.

With AI, each work minute feels more impactful than it did before. I can do more with every minute. And when it's easier to be productive, when it's easier to make money and create value with my time, it's incredibly incentivizing to put in more minutes and hours, not fewer.

That's the paradox. It's lovely to imagine AI enabling more time for life. In reality, it might incentivize us to work more.

That said, if you believe in the shift toward solopreneurship (which I do), here's where it gets interesting: I'm actually working fewer hours now, probably around 30 a week. Solopreneurship, controlling my own destiny, and AI together enable me to work less and spend more time on household admin, being a present parent, being with my kid.

I feel the trade-off deeply. I often wish I had more hours for work. But that's the key thing: I'm making this choice. I could work more hours and pay someone to help with the household and parenting logistics. It's all within my control.

Looking ahead

Will everyone embrace all of this immediately? No.

People still lean into doom and gloom when they think about AI. That's human nature. Everyone hates change. Everyone worries about job loss (which is a fair concern). I expect we'll continue to see instinctive, gut-negative reactions to AI developments throughout 2026 and beyond.

But give it five to ten years, and I think AI won't even be its own topic anymore. It'll just be embedded in how we live and work. Not a thing we debate, just a thing we use.

In the meantime, 2026 is going to be a fascinating year to watch and build in.

I love using this quiet time to think about the year ahead.

Usually, this involves a blank notebook and a lot of staring at the ceiling. But this year, I'm running a Personal AI Offsite.

Here is the agenda for my AI Planning Session (and how you can do it too).

1. The "Data-Driven" Look Back

We often rely on vibes to judge how our year went.

If you've been using AI tools this year, you have a goldmine of real, personal data. Let's hold up that mirror and get to the juicy insights.

(Note: I'm referencing specific tools I use, but you can do all of this with your go-to AI tools. It works best with the tool(s) that know you best.)

How to do it: If you use Granola (or any meeting recorder that allows you to query bulk transcripts), or if you have ChatGPT Memory enabled, run these prompts to get the truth about your year:

• The Time Audit: "Looking at my meetings/chats from 2025, what 3 topics consumed the most time? Did these align with my stated goals, or was I getting pulled into the weeds?"
• The Impact Audit: "What activities or decisions actually moved the needle on my [Revenue/Project Goals]? Contrast this with the activities that felt urgent but produced little result."
• The Energy Audit: "Analyze the sentiment of my voice/text. When did I sound the most excited and confident? When did I sound the most drained, frustrated, or repetitive?"
• The "Blocker" Audit: "What recurring problems or 'fires' did I spend the most time fighting? Is there a pattern to what keeps distracting me?"

You can go even deeper by querying your Google Drive, Calendar, or LinkedIn data.

2. The "Coach" (Strategic Planning)

Once I have the data, I switch to planning.

1. Start Without AI

Grab a notebook (or a blank doc) and brainstorm your goals for 2026. What do you want? What does success look like for you come December 2026? Write it down in your own words first. AI is great at refining, but only you know what your hopes and dreams are.

2. Go Deeper with AI

Now bring in the AI. Share what you wrote and let it help you expand and clarify.

"Here are my goals for 2026: [paste your notes]. Help me go deeper. Ask me questions to clarify my vision and surface anything I might be missing."

I don't like typing out long brain dumps, so I use Whispr Flow to dictate. I ramble about what I want, and let the AI structure it.

3. The Values Check

Before you go further, make sure your goals actually make sense together:

"Based on what I've told you, do my goals actually align with what I say matters most to me? Call out any contradictions or tensions."

4. Narrow, Focus, and Get Specific

If you're like me, you'll have too many goals at this point. This is where AI really shines, helping you narrow down and get tactical.

I like asking: "Which 20% of these goals would likely yield 80% of the impact? What should I actually prioritize?"

Once you've narrowed, have it help you focus and plan. For example:

• If you run a business: "My most important goal is [X revenue / X customers / X launches]. Help me reverse-engineer this into monthly and weekly targets. What specific activities need to happen each week to hit that number?"
• If you work in a company: "My most important goal is [promotion / launching X initiative / hitting X metric]. What are the key milestones I need to hit each quarter? What will my stakeholders actually care about?"

I ended up spending several hours on this part. The forced focusing and detailed planning was extremely helpful.

5. The Reality Check

Before you call it done, stress-test the plan:

• "What needs to be true for this plan to work? What am I assuming that might not hold?"
• "Imagine it's December 2026 and this plan failed. What went wrong?" (The Pre-Mortem)

You can then ask the AI for ideas on how to mitigate the risks.

3. Make It Visual

Resolutions are words. It helps to create a visual asset to make your goals more tangible and easier to track.

Here are three ways to make it real:

• The Visual Roadmap: "Create a visual roadmap for my 2026. Style it like a clean, modern infographic. Show quarterly milestones for [Goal A], [Goal B], and [Goal C] leading to my end-of-year vision."
• The Weekly Scorecard: "Based on my plan, create a simple weekly scorecard I can use to track whether I'm on pace. Include the 3-5 leading indicators that matter most, with specific target numbers. Make it something I could print and post on my wall."
• The Magazine Cover: "Generate a photorealistic image of me on the cover of Forbes Magazine in December 2026, having achieved [Big Goal]." It sounds silly, but can be great motivation and inspire clarity.

I'm mainly using Nano Banana or ChatGPT's image generator for these.

4. The Accountability System

A plan is useless if you forget it by February 1st.

You need systems that put your goals in front of your face and keep you on track.

Option 1: The Automated Check-In

Set up a simple automation to force a monthly review. You can use ChatGPT, Relay, or even a recurring calendar event with an AI link.

The Prompt (to draft your check-in agenda): "Create a 15-minute monthly strategy review agenda. Include: (1) scorecard review, am I hitting my weekly targets? (2) what worked vs. what didn't, (3) what's the one thing to focus on next month, (4) do I need to adjust the plan?"

Option 2: The Weekly 5-Minute Pulse

Block 5 minutes every Friday. Open your scorecard and answer three questions:

• Did I hit my numbers this week?
• What's the one thing that moved the needle most?
• What's the one thing I'm avoiding?

Log it somewhere, a note, a spreadsheet, a voice memo. The ritual matters more than the format.

Option 3: The "Back on Track" Prompt

You will fall off in February. Expect it. When it happens, use this:

"I set these goals in January: [paste goals]. It's now [Month] and I've fallen off track. Help me identify the smallest possible action I can take today to restart momentum."

Option 4: The Accountability Partner

AI is great, but sometimes you need a human. Find one person, a friend, a colleague, a coach, and share your scorecard with them. Set a 15-minute monthly call. The social pressure is worth more than any app.

That's the agenda for your 2026 AI Strategy Offsite. By the end of this session, you should have:

1. Insights: A data-backed view of how you spent 2025 (and what you need to stop doing).
2. Vision: A clear picture of what success looks like in December 2026.
3. The Plan: Specific weekly and monthly targets to get there.
4. The System: A built-in accountability loop so you don't lose the plot by Valentine's Day.

Bonus Round: 3 More Ways to Use AI for a Reset

Once the big strategy is done, here are three lighter ways to use AI to set yourself up for success.

1. The "Gratitude" Audit

It's easy to let relationships slip when you're busy. Use AI to surface the people you need to thank.

• The Prompt: Ask Granola (or whichever AI knows your collaborators best), "Who did I meet with this year who gave me great advice or energy, who's worthy of a gratitude note?"
• The Action: Ask AI to draft a template, but you write the actual note. Send 5 of these before next week. It starts your year on a high note of connection.

2. The "Systems" Reset

Reduce decision fatigue by templatizing the recurring stuff you do every week.

• The Prompt: "I write a weekly status update for my team every Friday. Based on what makes a good status update, create a reusable template I can fill in each week in under 5 minutes. Include sections for: wins, blockers, priorities for next week, and any asks."
• Bonus: Do the same for meeting agendas, project kickoffs, or 1:1 prep docs. Build the templates once, reuse forever.

3. The Learning List

Instead of doom-scrolling, curate your input for the new year.

• The Prompt: "I want to get smarter about [Topic A] and [Topic B] this year. Curate a syllabus of 6 newsletters and 6 podcasts that will teach me and keep my interest. Explain why each is worth my time."

I hope these ideas set you up for your most successful, fulfilling year yet.

Wishing you a wonderful 2026!

#1 The digital holiday card

I love holiday traditions. I do not love holiday logistics.

This year, I couldn't face a physical holiday card. Designing, printing, addressing, mailing, return-to-senders, it was all time and money I didn't want to spend.

So I tried a different approach: I made our family holiday card as a tiny website, with my favorite AI website builder, Lovable.

Card link: mueller-family-holiday-card.lovable.app

Here's the workflow I used:

1. Started in ChatGPT - Brain dumped ideas for what I wanted to include, the vibe, colors, etc. and asked ChatGPT for more ideas (it delivered!)
2. Asked ChatGPT to generate one thorough prompt that I could paste into Lovable
3. Moved to Lovable - Pasted prompt, got a full first draft
4. Refined & personalized - Tweaked spacing/colors/flow, swapped in real photos & details, added a few fun sections
5. Published + shared - I used the default Lovable link (fast, easy, free) and dropped it into an email to our family list

If you want to copy my card or make your own, here's the full guide including links: Lovable Holiday Card Guide

If you build one, hit reply and send it to me. I'd love to see what you make!

#2 Storybooks as gifts

Quick recap from last week, because it fits perfectly here: you can use an image model (I used Nano Banana) to create a personalized storybook starring your kid, your niece, your best friend's dog, whoever!

I made one for my daughter's neighborhood friend group. More details here and here.

#3 Family trivia night

This is my favorite "pass the time" use case.

Prompt: "I need trivia questions to play a fun trivia game with my family."

Then iterate a few times as needed:

• "These are too easy. Make them harder."
• "Ages are: [X, Y, Z]. Interests: [sports, pop culture, history, Disney, etc.]"

#4 Coloring sheets for the kids' table

Another easy win with an image model (Nano Banana or the new ChatGPT image generator).

Two ways to do it:

1. Photo-based: Upload a photo of your kid and prompt: "Turn this into a clean black-and-white coloring page with simple outlines, kid-friendly, lots of open space, no shading."
2. Interest-based: "Create a coloring page of a kid building a gingerbread house with a golden retriever. Simple outlines, lots of open space, no shading."

Print a stack. Put out crayons. Buy yourself a few minutes.

#5 Ridiculous family videos

For pure entertainment: open Sora or Veo3 (the video creator tools from ChatGPT and Google), give it a photo of someone in the family, and prompt a silly scenario:

• "Turn grandpa into a rapper performing on a tiny stage in the living room."
• "Make dad sing opera while cooking a turkey."
• "Make my sister narrate a dramatic movie trailer about opening presents."

Do a few rounds and you'll end up with a full set of absurd clips that makes everyone laugh.

I hope these ideas bring some joy to your holiday season. If you try any of these, hit reply and tell me what you made.

Wishing you very happy holidays,

Mollie

Most people view meetings as a necessary evil, a time cost that keeps them from doing "real work."

But what if the meeting was the work?

This week, I want to share the highest-leverage workflow currently running in my business. It's a mindset shift that transforms a 30-minute Zoom call from a calendar block into a content engine.

The Problem: The "Post-Meeting" Slump

Usually, you finish a great call with a client or team member. You have incredible ideas, clear decisions, and a lot of energy. Then you hang up.

Now comes the hard part: You have to write the follow-up email, the proposal, the action plan, and ideally the LinkedIn post about the insight you just shared. Suddenly, that 30-minute meeting has created three hours of manual writing work.

The Solution: The Transcript as "Draft Zero"

The most efficient way to write is to speak. Humans can speak roughly 150 words per minute, but most of us type significantly slower than that.

When you record and transcribe a meeting, you aren't just creating a record; you are generating thousands of words of raw material. I no longer look at a transcript as a simple record. I look at it as a Draft Zero for almost every asset my business needs.

The Workflow

Here is the process I use to turn one conversation into multiple deliverables without typing from scratch:

• Record & Transcribe: I use Granola to capture the full transcript from my meetings.
• Identify the Asset: I decide what I need (Proposal? Workshop outline? Blog post?).
• The Single Prompt: I paste the transcript into my LLM of choice with one specific instruction to reformat that raw text into a polished final product.

Because the AI has the full context of the conversation, the specific details discussed, the client's objections, the agreed upon plan, the output is 90% ready to go.

The Best Holiday Gift: The Neighborhood Squad Book

Now, let's switch gears. We use AI to save time at work so we can do magical things in our personal lives.

For the holidays this year, I decided to create a custom children's book for my daughter and her four best friends in the neighborhood.

I used Gemini to write a story where all five kids (by name) go on an adventure through our specific neighborhood, and I used Nano Banana to generate consistent character illustrations for each of them.

How I did it

Here is the exact workflow I used to make this happen:

• The Script (Gemini): I told Gemini I wanted an adventure story set in our neighborhood starring five toddlers. I fed it their names and one defining quirk for each child to personalize the plot. It took time to get the plot to a version I liked.
• The Art (Nano Banana Pro): I then focused on nailing the "fairytale-esque" style for the first image. For every subsequent page, I instructed the AI to maintain that exact style and character set while referencing the specific text for that scene.
• The Assembly (Shutterfly): Once I was happy with the script and the visuals, I dropped them into a standard photo book layout and hit publish.

It is a beautiful gift, so personalized, and a true keepsake.

It did take time to create, about six hours, because I was being meticulous with the details. But it is relatively affordable ($30 per hardcover book on Shutterfly).

Know another parent who needs a holiday gift miracle? Forward this email to them so they can be the hero of their neighborhood, too.

And if you end up making one, send me a screenshot! I'd love to see what you create, and I'm happy to share more about mine.

Google just released Gemini 3.0, a significant model update. But the part that has everyone talking, and the part that changes things for us, is the upgrade to their image generator.

Officially, it's called "Gemini 3 Pro Image."

Unofficially? The internet has named it 🍌 Nano Banana.

Here is why this is a massive unlock for non-technical people like us who just want to create, save time, and build businesses, and the 5 specific ways I'm using it this week.

What is "Nano Banana"?

Before getting into the use cases, it helps to know why the name stuck. It's a mix of viral marketing and internet culture.

• The Original: The previous model was codenamed "Nano Banana" internally. The name leaked, the internet loved it, and Google embraced it, even adding a banana emoji (🍌) to the prompt bar.
• The New Version: The update is officially "Gemini 3 Pro Image," but everyone, including tech reviewers and Google's own marketing, is calling it Nano Banana Pro.

Why is it a game changer?

Unlike previous AI image generators (like Midjourney or older DALL-E versions) that just "dreamed" up pixels based on vibes, Nano Banana Pro uses the Gemini 3.0 language model to "think" before it draws.

Here is the breakdown of why that matters:

• It Can Finally Read and Write: This is the biggest improvement. Previous AIs would turn text into alien gibberish. Nano Banana Pro can generate perfect text on signs, t-shirts, wedding invitations, and logos.
• It Has "Common Sense" (Reasoning): Because it uses a "thinking" model (LLM) to plan the image first, it understands logic. If you ask for "three red balls and one blue cube," it actually counts the objects and places them correctly. It can even solve visual math equations.
• Real-World Grounding: It is connected to Google Search. You can ask it to "Draw an infographic of the current weather in Seattle," and it will pull the actual weather data and generate a correct graphic.
• Character Consistency: It can "remember" a person's face. You can generate a character and then put that exact same person in 10 different scenarios (eating, running, sleeping) without their face morphing into a stranger's.

5 ways I'm using this (and you can too)

This isn't just about making pretty art. It's about workflow. Here are the specific use cases relevant to what we do.

1. One-Shot Branding & Animation

I wanted a new logo for Watch Me AI. Nano Banana nailed it. Here was my initial prompt:

And here are the options I got to in a few minutes:

You can vote for your favorite here!

I took it a step further. The embedded video creation feature allows you to take static assets, like your company logo, a family crest, or even a drawing your kid made, and animate them in a fun, magical way.

Watch the animation video

My prompt for this animation was simply: "can you animate it so we see the sparkles sparkling?"

2. The clearest instruction manuals

This is one of the coolest functional use cases I've seen online. You can upload a photo of a physical object and ask the AI to overlay instructions on it.

• The Example: Someone took a photo of the inside of their car hood and prompted: "Add instructions to this photo of how to change the air filter."
• The Result: The AI drew arrows to the exact latches and parts, with text labels explaining the process.

Think about the applications for this: troubleshooting something broken in your house, explaining a setup to a client, or creating quick how-to guides.

3. Incredible visual summaries of complicated information

We are moving beyond simple images into "Data Storytelling."

I saw a great example from Eric Sun, who created a one-shot infographic for the Acquired Podcast's 3.5-hour episode on Trader Joe's. He fed the AI the context of the episode, and it synthesized hours of audio content into one detailed, accurate visual map.

I'm guessing his one-shot prompt was something like:

"Here is the transcript/summary of the Acquired Podcast episode about Trader Joe's business model. Synthesize this entire episode and summarize it as a clear visual diagram in Trader Joe's branding"

For what it's worth, I still use Napkin.ai (featured in last week's newsletter) for clean, specific visual diagrams (it's still top-tier for that), but for synthesizing massive amounts of messy content into one visual story, Nano Banana is a beast.

4. The Course Creator Workflow (Transcript -> Visuals)

I'm currently recording modules for my new course. My usual process is to speak through an outline to find my flow, then record the final version.

With Nano Banana, I've added a step that saves me hours of design time.

My prompt:

"Here's my script for a module of my AI for leaders course im recording. Can you turn this into a few great infographics i can share on screen as i talk? Please create in my brand colors, which are in the attached image."

Because the model understands the logic of the text, it creates slides that actually match what I'm saying.

5. Visual Explainer Breakdowns

Because the model has "common sense," it can reconstruct scenes. People are using it to break down how movie scenes were filmed, asking it to draw the lighting setup and camera angles. It infers the technical details from the final image and draws the "behind the scenes" diagram.

Summary: The Upgrade at a Glance

If you're skimming, here is the difference between what we used to use and what we have now:

This technology is moving fast. My advice? Pick one of the use cases above, maybe the "instruction manual" photo or the slide creation, and try it out this week.

See you next time (unless the internet shifts again).

Mollie

The leadership AI divide

Most senior leaders I talk to today fall into one of two camps:

1. Those who delegate AI adoption, setting strategy, assigning projects, and checking in on progress.
2. Those who use AI themselves, testing tools, experimenting, and integrating it into their daily rhythm.

In a new series we're launching called Watch Me AI for Leaders, Nick and I explored why that second group is quietly pulling ahead, and how hands-on leaders build sharper instincts, make faster decisions, and earn more credibility.

Key insights from our conversation

Here's what stood out most from our 15-minute chat:

Credibility comes from experience. Nick shared a story from his time leading culture and talent at Ørsted. His team was using AI for interview design, but he realized later that without hands-on experience, he couldn't challenge or connect the work to broader strategy. Now that he uses AI daily, he feels more curious, confident, and capable of leading transformation.

AI isn't a task to delegate, it's a tool to practice. Leaders who get involved directly discover where AI truly accelerates work: decision-making, communication, and reflection.

The biggest blockers are time and mindset. Leaders often assume AI is too technical or time-consuming to learn. In reality, it's fast, intuitive, and immediately useful. Even five minutes before a meeting.

How to start using AI as a leader (from our conversation)

These are just a few of the practical ways we cover in the episode for integrating AI into your day-to-day as a leader.

1. Before meetings: Ask ChatGPT, "Help me clarify the goal of this meeting and where we might get stuck."
2. After meetings: Use AI note takers like Granola or Fathom to summarize key points and action items, draft follow-up emails, and even create presentations.
3. When facing a decision: Prompt, "Help me structure my thinking, what's the core decision and what scenarios should I weigh?"

These small, repeatable habits build your confidence and speed over time.

Watch or listen

Watch the full episode (15 min): "Should Leaders Actually Use AI?" → YouTube

You'll hear us talk about:

• Why credibility matters more than ever
• Where to start if you feel "too busy" to learn AI
• The mindset shift from AI strategy to AI practice

AI Bits & Bobs

Content Series: Portraits of AI Across America

A collection of stories about how people in all 50 states are using AI at work, authored by a fellow IDEO alumnus, Jessica Herman, and published by Google. While a bit heavy on the Google mentions, it's a great collection of diverse AI use cases. In this exploratory era, we're all learning how to use AI by seeing others use it in novel ways, so content like this helps broaden usage.

I particularly liked this story about a Montana digital learning academy using AI to scale education to the state's most rural areas.

Create a jingle with one prompt

I used Sora 2 to create a 10-second intro jingle for YouTube videos (the music at the start of this video). Here's the prompt I used (written by ChatGPT):

Create a 10-second jingle for a podcast called "Watch Me AI." The sound should feel modern, confident, and intelligent, with an upbeat tech vibe that evokes creativity and innovation.

Include: Bright synth tones, light percussion, and a subtle digital texture; A clean ending (no fade-out); Tempo: around 120 BPM; Genre reference: somewhere between electronic pop and startup-intro music; Mood: inspiring, future-focused, slightly playful but professional.

It gave me a video when I only needed an audio file, but I just hid the video behind an image in Canva (where I edit videos), so you can't even tell. I could have used a fit-for-purpose AI music generator, but I already pay for Sora.

Financial forecasting prompt

I updated my financial plan this morning. Here's how:

1. Did some really basic outlining of how Watch Me AI makes money. Basically just wrote down my revenue streams, current pricing, and a guess at volume for 2026.
2. Pasted a screenshot of that spreadsheet into ChatGPT and said: "Here's some quick math on my revenue streams. Can you review this for reality?"

Here's what it gave me back:

• A reality check on every assumption
• Estimated income ranges for each revenue stream based on real benchmarks
• Analysis of my time and energy investment by line of business
• A recommended focus plan for 2026

In about 30 minutes, I had a clear picture of which parts of my business are the highest leverage and which ones don't scale as well. It helped me refocus my energy toward the work that's both more lucrative and more energizing for next year.

The "tech bro" alter ego that raised my game

When I was fundraising for my startup, I spent a long time working on my pitch deck. I got it to "good", but I'd been through this process before, and I knew that when I leaned into my alter ego -- an insanely confident, overly self-assured, full-of-swagger San Francisco tech bro -- I performed better.

So this time, I tapped ChatGPT for help. I uploaded my deck and gave it this prompt:

"Rewrite this pitch deck in the voice of an insanely confident tech bro who's raised billions of dollars and is barely deigning to speak to investors."

The new version was bold, direct, and unapologetic. It captured the conviction I had in the language investors respond to. And it worked. I got more investor meetings, more second meetings, and I felt way more confident walking into every room.

That was one of the first times I realized how powerful AI can be for confidence.

How AI builds real-world confidence

AI makes it possible to do things you've never known how to do, and to do them well enough to move fast.

Before this wave of generative AI, I couldn't write code. I could design and strategize, but I couldn't build digital products. Tools like Lovable and Claude Code changed that. Now, I can describe what I want in plain English ("build a weekly win tracker that pulls from my meeting notes") and have a working product in hours.

That shift alone made me feel competent, capable, and even unstoppable.

In addition to product building, I'm constantly needing to learn new things: marketing strategy, pricing models, course launches. I can now say, "You're my marketing strategist. Here are my goals -- build me a plan."

Instead of feeling blocked by what I don't know, I can get a smart, confident first draft in minutes and iterate from there.

And it's not just founders. AI gives that same confidence boost to anyone navigating modern work.

Recording my meetings and being able to instantly generate follow-up emails or slide decks makes me feel confident that nothing slips through the cracks. I can respond faster, stay organized, and turn deliverables around without losing momentum.

And when I can create beautiful, on-brand slides myself -- using tools like Canva's AI-powered slide designer -- I don't need to wait on design support or schedule around others. That flexibility gives me more control over my time and the confidence to ship what I want, when I want.

Beyond productivity, AI has become a coach too.

It helps me prep for meetings, reframe tricky conversations, and even pep-talk myself into a stronger mindset. When I need to ask for more money, get buy-in, or plan my week around my goals, I turn to ChatGPT. It doesn't judge, it's always available, and it helps me think clearly and act with confidence.

Try This Yourself

Here are a few ways to start building that same confidence with AI this week:

1. Step into a stronger voice
"Rewrite this email/deck/pitch in the voice of a leader who's clear, confident, and persuasive."

2. Level up your communication
"Edit this to sound polished and executive-ready -- direct, warm, and easy to follow."

3. Prepare for high-stakes moments
"You're my executive coach. I have a big [presentation / conversation / review] coming up. Help me get centered, highlight my key points, and deliver with conviction."

4. Confidence through clarity
"I'm tackling [task or project]. Break down what great looks like and help me make a plan that feels achievable."

5. Design your best week
"Help me visualize my ideal week -- where my time, energy, and priorities are aligned with how I want to lead."

AI Bits & Bobs

Lovable x Shopify: Lovable just launched a Shopify integration, which means you can now spin up a full online store from one prompt. If you've got a craft you've been itching to monetize, it's literally never been easier.

ChatGPT Atlas: OpenAI released Atlas, its new browser. I've just started using it. So far, I love being able to chat directly in the browser about whatever I'm working on -- like writing this newsletter. I'll share more as I go deeper.

We're all constantly networking, for jobs, partnerships, opportunities, and the difference between a dead end and a warm intro often comes down to how you write the ask. Most people don't do it right, and for most of my career, neither did I.

A few years ago, I hired a very smart advisor (hi Arjun Dev Arora!) who taught me how to do it the right way. He calls it a Request for Introduction (an RFI), and I've come to think of it as a "forwardable" email. Because that's the real secret: the email has to be forwardable.

Why "forwardable" is the key when asking for an introduction

When you ask someone for an introduction, you're asking for a favor. You're asking them to take time out of their day and use their social capital to help you get something you want.

So here's the number one rule: make it as easy as possible for them to do this favor.

A forwardable email means your contact can literally hit "forward," add a quick note, and send it. They don't need to delete or rewrite anything. The entire email, including the part addressed to them, is designed to be forwarded as-is.

That's what makes it work. You've already given them the context, the "ask," and a short blurb about you that makes sense to the recipient. So they feel comfortable using their social capital to do you this favor, and they don't need to spend time writing their own email.

The second rule for RFIs: make it as easy as possible for the recipient to say yes.

How do you do that? Clarity, personalization, and a touch of flattery. Let's break it down.

The anatomy of a great Request for Introduction email (RFI)

Here's the structure I use every time (with examples below).

• Subject line: Simple and clear: Mollie Mueller ↔ [Name] Intro
• Greeting: Address it to the person you're asking to make the intro. Keep it short and warm.
• The Ask: Clearly ask for the introduction, and write one to two sentences on why you want the intro to them. It should sound thoughtful and slightly flattering toward the person you're trying to meet.
• The sign-off: A mention of your blurb below, and a thank you.
• The blurb below: A concise, strategic, personalized blurb summarizing you or your company. Include:
  • A short "about me" or "about [company]" paragraph, one or two sentences summarizing who you are.
  • 3-4 bullet points, highlights that make you relevant or credible.
  • Links, add your LinkedIn or website if helpful (and make sure they're also in your email signature).

The goal: make it clear why you're worth meeting and make it effortless for the forwarder. They should send the entire email to give the recipient the full context.

Here's an example of a job hunting RFI:

In the above example, Alex (the sender) has made a clear ask, given a strong reason for "why," added details/flattery on "why them" (i.e., specific mention of Jordan's relevant experience), and included a personalized "why me" section to prove relevance.

In this case, Taylor (the introducer) should hit forward, add a short note like "Hey Jordan, Alex is one of the best PMs I've worked with, I'd love to introduce you" and hit send.

When to send an RFI

The most common reasons you need an introduction are for job hunting and sales. But this format works for any introduction.

For job seekers

When you're applying for roles, a warm introduction can make or break your chances. Without one, your resume is just one of thousands in an applicant tracking system.

But remember: people at top companies get a lot of these requests. You need to make it easy for them to say yes. Do your homework. Show that you've paid attention.

For example:

• Reference an article or post they've written that inspired you.
• Mention a shared alma mater or professional interest.

Then, in your "about me" section, make your bullet points directly relevant to the role or team. Show you've thought about why you belong there.

For founders, sellers, or solopreneurs

If you're in sales, fundraising, or partnerships, the same logic applies. You're asking busy people to connect you with potential customers, investors, or collaborators. Most have never heard of you and may not care yet.

So do the work for them:

• Be specific about why you want to meet this person.
• Show that you understand their world.
• Make it obvious how the connection could be mutually beneficial.

Here's an example of a sales RFI:

A note for founders and fundraisers

If you're raising money, the RFI format is essential. You'll likely ask dozens of people for introductions to investors, and this is the fastest, most respectful way to do it.

How AI can help

What I've described so far takes real effort, and it should. But AI can now handle much of the tedious part: the research and customization.

Before AI, I'd make a target list, open every LinkedIn profile, read bios and posts, find that "why them" nugget, and tweak each "about me" section manually.

Now, I let AI do 80% of that work.

Here's the simplest process:

Open up ChatGPT, Claude, or your favorite general AI tool.

Feed it:

• Your RFI template (you can use mine below)
• The name and LinkedIn URL of the person you want to connect with
• Context about your goal (job, partnership, etc.)

Ask it to:

• Research that person
• Suggest a relevant reason for why them. (You're likely clear on why that company, but you need to personalize it to the human)
• Customize your "about me" section or bullets to match your goal and their interests

If you're job-hunting, you can paste in the job description so your bullets are tailored to that role.

RFI Template (copy and adapt)

Subject: Mollie Mueller ↔ [Person you want to meet's name] Intro

Hi [Name],

Great catching up last week. Could you connect me to [Person]?

I'd love to meet them because [short, thoughtful reason].

I've included a quick blurb below. Thank you so much.

Mollie

–––

About me:

I'm [1-2 sentence summary].

Highlights:

• [Bullet 1: relevant achievement or role]

• [Bullet 2: notable project or result]

• [Bullet 3: credibility or shared connection]

Links:

[LinkedIn] | [Website]

AI Bits & Bobs

Tool to try: Napkin AI

If you've ever sketched an idea on paper and wished it could magically become a clean, professional diagram, this one's for you.

Napkin AI turns written text into editable, branded diagrams that actually look good. You can write a short description ("org chart showing product, design, and engineering under CEO"), and Napkin instantly generates polished visuals.

Why I like it:

• Transforms memos or notes into visuals that communicate complex ideas clearly
• Lets you iterate: tweak text or nodes and regenerate in seconds
• Exports easily to Google Slides or Docs for further editing
• Offers affordable plans (including a free tier and $9/month mid-range)
• Unlike Gamma or other slide tools, Napkin's diagrams don't look templated, they look designed

If your work involves explaining systems, workflows, or org structures, this is the best tool I've found for making that process fast and beautiful. Let me know if you try it and what you think.

Insight: 75% of enterprises are already seeing ROI from AI

Wharton's 2025 AI Adoption Report found that three-fourths of enterprises report a positive return on their AI investments.

There's been a lot of noise from skeptics claiming AI hasn't proven its ROI yet. Those of us working in it see the returns daily but until recently, they weren't being measured effectively. I expect 2026 will be the year of AI ROI, with more companies moving from experimentation to clear measurement and impact tracking, and this early research is promising.

Read the full Wharton report →

P.S. If you've used AI to personalize outreach or intros, reply and tell me what worked. I might feature it next week.

If you only try one new AI thing this week, make it voice mode.

Storytime: Ryan's most productive commute

Ryan is a longtime friend and collaborator of mine. He's a performance coach who works with corporations and senior executives, and he recently told me a story about how he's using AI that I had to share.

He had a big meeting coming up with a retail client and was about to drive an hour to get there. Before leaving, he dropped his analysis pack into ChatGPT. His plan: use the drive to think through the material and sharpen his strategy.

So that's what he did. With ChatGPT voice mode on, he started talking through the entire analysis while driving, asking questions, pressure-testing his assumptions, and thinking out loud about the storyline he wanted the client to hear. ChatGPT didn't just reflect his words back; it challenged him to prioritize what really mattered, highlight patterns across the data, and reframe points in clearer, more persuasive language. It became a sparring partner, helping him move from "a bunch of findings" to "a tight narrative with a clear takeaway."

When he pulled into the parking lot, he gave it one more prompt: "Format this so I can drop it into Gamma when I park." ChatGPT reorganized everything into clean, slide-ready chunks. Ryan hotspotted his phone, pasted it into Gamma, and within minutes had a polished deck ready to go.

In the meeting, he didn't lean on the slides. He said, "I've got some key points on slides, but I really just want to tell you the story." And it worked, the client was engaged, the conversation flowed, and Ryan felt confident.

Later, he told me: "That one experience completely changed how I approach prep work. I realized my first pass should always be a conversation with AI (ideally while I'm driving). It's not about skipping prep. It's about using time I already have to prepare smarter."

How to Make Voice Your Productivity Superpower

I love voice mode because it lets me capture messy thoughts quickly without needing to organize or structure them (yet). It removes friction from getting started, forces clarity, surfaces blind spots, and gets me outside.

Let's talk tactics.

When to Use Voice Mode

• Car rides - My commute becomes prep time or strategic thinking
• Dog walks - I find I process thoughts best while getting fresh air
• Household chores - Makes folding laundry feel so much more valuable
• Sitting at my desk - Sometimes, even when I'm at my computer, it feels more natural to talk it out

The Tools

1. ChatGPT Voice Mode is the main player here. It's conversational, responsive, and gets smarter the more you use it. Ryan used it to work through his meeting prep, then asked for a specific output: "Format this so I can easily drop it into Gamma when I park." That prompt turned an hour of thinking into a ready-to-use deck.
2. Voice Notes (on your phone) are your other secret weapon. Unlike ChatGPT voice mode, voice notes don't respond; they just capture. This is perfect when thoughts are coming fast and you don't want interruption or the need to speak coherently (yet). Capture everything now, have AI organize it later.
3. Gamma is what Ryan used to turn his formatted content into slides instantly. It's the leader in AI slide creation. Gamma.ai
4. "Connectors" in ChatGPT: I have my Gmail, Google Calendar, and Google Drive connected to ChatGPT, which means the AI can pull context from my actual work life. Here's how I use this on the way to meetings:
   • If I'm meeting someone I've worked with before, I'll say: "Summarize my latest emails with [name] and pull up notes from our last meeting." The AI gives me a quick brief so I'm not walking in cold.
   • If it's someone new, I'll ask: "Look up [name] on LinkedIn and tell me about them." By the time I arrive, I know their background and can have a more informed conversation.

This feels like having a personal assistant who's already read everything and is ready to brief you.

Try This Yourself

Pick one of these to try this week:

For Meeting Prep

Talk through an upcoming meeting with ChatGPT voice mode and have it coach you on setting goals, structuring the meeting, and starting with a strong opener.

Prompts:

• "You are my strategic sparring partner. I'm prepping for a meeting with [Client]. Here's the analysis pack [pasted/attached]. Help me: (1) extract 3 executive takeaways, (2) map data to story beats, (3) anticipate 5 stakeholder objections with responses."
• "From the above, force-rank what matters to the client by impact times proof. If something is 'interesting but not decisive,' say so and park it."
• "Reformat into slide-ready chunks: Title, 1-sentence takeaway, 3 bullets (evidence), and 'So what?' for each."
• "Draft a 60-second opening that frames the problem, stakes, and what we'll decide."

For Brainstorming

Open a voice note and start rambling about something that's been on your mind. Maybe a podcast or article stuck with you, and you have reflections. Maybe your kids just started school and you're pondering new beginnings or the joys of youth. Whatever it is, just talk it out. This could become a blog post, a thoughtful letter, a journal entry, or a LinkedIn post.

Later, paste the transcript into ChatGPT with one of these prompts:

• "Turn this into something coherent"
• "Turn this into a to-do list with priorities"
• "Turn this into a blog outline"
• "Extract the three most important ideas here"

For Daily Clarity

On your morning walk or commute, open ChatGPT voice mode and talk through your day.

Prompt: "Help me think through my priorities for today. I have [meetings/projects/tasks]. What should I focus on first?"

Need help finding these tools? Ask ChatGPT (or Claude, Perplexity, etc.). Say "how do I start voice mode?" or "how do I find the voice recorder app on my phone?"

One of my favorite ways to use ChatGPT is for financial planning. It can turn your budget and dream goals into a roadmap.

Try this:

Given the details below, help me understand how much money my partner and I need to earn (and save) to achieve our dream life in [10] years.

• Regular household expenses (I took a screenshot of our budget spreadsheet and pasted it into ChatGPT).
• Dream life: a list of "dream life" expenses, like a vacation house, with rough cost estimates. We do this activity once a year as a couple.
• Current income for both of us and estimates of growth.
• Other life details like # of kids/dependents, location, etc.

Please make reasonable assumptions for tax rates, savings growth, inflation, etc.

Disclaimer: This will be directionally accurate and needs refining to turn into a plausible roadmap. Don't make any big life decisions based on the first round of results!

Work Hack I Recommend

Spend 5 minutes prepping with ChatGPT before every meeting. You will have more productive meetings and greater impact.

My simple prep process:

• Start with this prompt: "Help me prep for my upcoming meeting with [name]. Here's the context: [copy/paste emails or screenshot how this meeting got set up]. My goals for this meeting are: [totally fine to say you don't know yet]"

  Stop there if you're short on time. You'll get solid advice from just that.

• Go deeper (when it matters): "Research [name] and help me predict what they'll be interested in. Strategize how to use these 30 minutes to build towards my goal of ___."

  This is where it gets interesting. ChatGPT will often surface things about their background, recent projects, or company priorities that I missed, and help me better shape the conversation towards shared goals.

• Fill knowledge gaps: If they work in an industry I don't know well, I'll say: "Explain [their company/role/industry] to me like I'm smart but new to this space."

I included a quick example in this demo video.

Deep-ish Thoughts

One of my big predictions for the future of work is the rise of rampant entrepreneurship. AI has already removed the technical barrier, anyone can now spin up a site, product, or agent without knowing how to code. Now we're watching the growth barrier fall too. Lovable, my favorite vibe-coding platform, just made its customers' business growth its north star metric. That's a rare move, and it signals a shift toward tools that don't just let you create, but actually help you build traction.

Read the post here: The Growth Barrier Falls: How Lovable's Metric Shift Unlocks the Next Phase of Rampant Entrepreneurship

Just for Fun

I've been trying out Google's new image editor tool called Nano Banana. Here are a few recent explorations for your enjoyment.

My goldendoodle is a notorious food thief. Impossible to capture on camera, I used Nano Banana to re-create her recent heists.

It's planting season here in the PNW and I needed inspiration for a boring patch of dirt. Nano Banana delivered an excellent mockup.

Longer video with more non-work-related use cases for Nano Banana, plus instructions, here: YouTube

As someone who lives and breathes applied AI, here's something most people miss:

Yes, today's AI tools make mistakes. They hallucinate, they spit out sloppy drafts, they build landing pages where the boxes don't quite line up, they generate images that look slightly uncanny. They need too much prompting to get to "good enough." They are, in many cases, not as good as humans at the basics.

Most people stop there. They try an AI tool, get frustrated by the rough edges, and walk away with the conclusion: "This isn't good."

But here's the reality that's underestimated: AI improves at a pace that's almost impossible to wrap your head around. Unlike most consumer tech, where major upgrades come annually, these systems get better weekly. The Stanford AI Index reported that model performance on standard benchmarks is now doubling in less than a year. In practical terms: a tool you're disappointed in today may feel like a completely different product 6 weeks from now.

The problem is when tools overpromise their current state. I've experienced this recently with Lindy, an AI agent builder that announced you could "build an agent by just describing what you want." Twice now, I've felt suckered, not because the tool was clunky (that's normal), but because the promise set an expectation the current reality couldn't meet. I ended up in the usual back-and-forth debugging, which would have been fine if I'd expected it.

This creates a dangerous dynamic. When people's first AI experience involves feeling overhyped and disappointed, it becomes their formative impression of all AI tools. They're less likely to keep engaging, which means they miss the actual capabilities developing right under their noses.

Compare this to tools that set realistic expectations. When Lovable, Relay.app, or Granola announce features, they deliver what they promise. The tools might still be clunky, but I'm not primed for disappointment. Instead, I'm calibrated to appreciate the steady improvements.

I've seen this with my own work. When I first started using Granola seven months ago, the chat feature wasn't great. I'd ask it to write a meeting follow-up email and its output was way worse than ChatGPT, so I'd copy-paste the transcript into GPT. As of a few months ago, Granola's chat got WAY better. It now writes solid post-meeting emails and gives excellent responses to questions like "Help me prep for the follow-up meeting with this person."

The key insight: evaluate tools by their improvement trajectory, not their current flaws. Those of us building with AI learn to tolerate quirks because we know they won't be quirks for long. The average consumer evaluates the tool in front of them. Builders evaluate the curve of improvement behind it and the acceleration ahead.

AI tip of the week: ask Granola questions across ALL your meeting notes

If you've been following me, you'll know I LOVE Granola, the AI meeting note taker. Here's my new-ish favorite thing to do in Granola: ask it questions about ALL of my notes.

Example: I recently got back from holiday and had the classic "What do I do for work again?" foggy Monday morning start. I opened up Granola and asked all four of its recommended prompts (see image). Immediately felt more back in it.

Note: You might want to change the setting at the bottom to "All" meetings to get the full picture.

New Resource: "What should I vibe code?"

I'm super excited about vibe coding (surprise surprise) and want everyone to benefit from it, but I hear a lot of people say, "I don't really know how to use this for my work".

So I built this simple tool. You put in your job title or other descriptor of your work and get ideas + prompts for things you could use vibe coding to achieve.

Excited for y'all to try it and give me feedback. Eventually I'll have to charge for this (those GPU's ain't free!) but for now it's open for testing.

Try it here: What Should I Vibe Code?

Quick pulse: Nano Banana

Have you tried Nano Banana (Google's new image generator)? Lots of hype, but I haven't found my winning use cases yet. If you have, reply with what worked! I'd love to hear.

I've been experimenting with GPT-5 and it's definitely a different world of prompting than before. Here are a few tips:

• Give longer, more detailed prompts. Pre-GPT5, short prompts worked. Now, the more detail the better. I like to ask it to give me the full prompt.
  • Example: "I want to create a how-to guide for LinkedIn content agent building based on the recording from my live session and other stuff I've done. What's a good prompt to get you to create a really great guide?"
• Literally tell it to "think harder." (Write those two words in your prompt). This has been surprisingly effective.
• I'm loving its Next Steps. One thing that's been consistently excellent is the Next Steps GPT-5 recommends at the bottom of its response. I find myself often saying "yes" to those.
• It's amazing with tools ("connectors"). I needed an updated list of prior Crew coaches. The info lived across a few old Google Sheets + emails. I asked GPT-5 for an updated list, and it pulled the Sheets + recent emails together into one clean list I could copy into my outreach email. Here are my exact prompts:
  • First ask: "I'm looking for a full list of all former Crew coaches. These are people I would have emailed as a group several times in 2024 and a few times in early 2025. Are you able to find that list for me from my Gmail?"
  • Follow up: "Great start. There should also be a longer list of coaches, maybe in my Google Drive, including coaches who had applied to Crew but didn't end up coaching with us. See that anywhere?"
  • Note: You'll need to connect your ChatGPT to your email and G-drive (or other file storage system). You can do this in Settings, Connectors.

Recap: build a LinkedIn content agent with me

Last week's live session focused on LinkedIn content creation (that thing we all hate but know we need to do). We covered how to:

• Automate your LinkedIn content pipeline
• Use AI agents to research, write, and schedule posts
• Build workflows that still keep your authentic voice

AI tip of the week: trivia games on demand

A fun use case: I was with my extended family the other day, waiting for food at a restaurant. We were hangry and in need of a distraction, so I asked ChatGPT for a trivia game. It came back with a list of questions. I skimmed them, decided they were too easy, told it to make them harder, and ended up with a perfect game that killed the hungry wait time.

My lazy prompt: "Give me a trivia game for my family while we wait for food."

Try it next time you need a quick group activity. It works great across generations.

My deep thoughts this week

Lots of deep thoughts but no new essays this week, I've been on vacation.

"I know I need to start using AI... but I don't even know where to start."

I heard this exact phrase from a friend last week.

If that sounds like you, smart, ambitious, and feeling a bit behind on the AI curve, you're not alone. The good news? You don't need to read another lengthy article or take an expensive course. You just need to start doing.

Here's your roadmap to AI mastery, broken down into three practical phases that will take you from overwhelmed to empowered.

Phase 1: Dip Your Toes In (Start This Week)

1. Start Recording Your Meetings

Use a free tool like Granola (or another AI note-taker) and let it run in the background during your calls. You'll get automated transcripts, follow-up emails, and action items with zero effort. It's one of the fastest ways to see the tangible value of AI in your day-to-day workflow.

2. Browse Templates on Relay

Think of Relay.app as your intro to AI agents. Even if you don't build anything yet, just scrolling through their templates will give you a 60-second crash course in real-life AI use cases. Perfect for busy professionals, operators, marketers, you name it. If something resonates, you can launch it with a single click.

3. Ask ChatGPT Anything

Literally ask:

• "How should I be using AI for my business/job?"
• "What are 3 tools I should know about?"
• "What would you do in my shoes?"

The more you ask, the more natural it becomes. AI isn't just a tool. It's your thinking partner. And the only way to get better is to use it.

Phase 2: Get Creative and Connected

Once you've conquered Phase 1, you're ready to expand your AI toolkit:

1. Experiment with Canva's Image Generator: Dream Lab

AI image generation has come a long way. Pick a fun prompt like "a cat wearing a business suit giving a presentation to a room full of dogs" and experiment with guiding the AI on style. Try adding specific style instructions like "in the style of a vintage poster" or "minimalist line art." Upload a reference image to see how the AI adapts. It's surprisingly powerful for creating marketing materials, social content, and brainstorming visuals.

2. Follow AI Influencers and Content Creators

Seeing how other people use AI will give you loads of inspiration. Ask ChatGPT for recommendations on who to follow based on your interests.

Here are a few of my favorites:

• Me, (obviously!): LinkedIn. Instagram.
• Jacob Bank
• Saanya Ojha
• Ethan Mollick
• Felix Haas
• For audio content, check out Nathaniel Whittemore's podcast "AI Daily Breakdown" for quick, digestible AI updates.

3. Use ChatGPT as Your Personal Coach

This might feel weird at first, but try asking ChatGPT questions like:

• "How do I prioritize my tasks when everything feels urgent?"
• "What's holding me back from launching this idea?"
• "How do I handle impostor syndrome in my career?"
• "Based on what you know about me, what 3 areas would you prioritize for my growth?"
• "Based on our conversations, what patterns do you notice in my thinking?"

You'll be surprised how thoughtful and helpful the responses can be, especially as ChatGPT learns more about your specific challenges and goals.

Phase 3: Go Pro

Ready to level up? Here's where things get exciting:

1. Build a Simple Agent

Go back to Relay, or try Zapier or Gumloop (they're all similar). Start with a template and walk through the guided steps to build your first simple agent. This is where you'll start to see the real power of AI automation.

2. Get an AI Companion (Try Tolan)

This will feel silly at first, the product is aimed at young people, but it will open your eyes to the future of AI companions. It's a glimpse into how AI relationships might evolve and how they could support us in unexpected ways.

3. Build with Lovable

You'll see how easy it is to build a website, and if you care to go further, it'll guide you through building elements that turn your site into more of a product. Perfect for testing ideas quickly.

Your AI Journey Starts Now

The beauty of this approach is that each phase builds on the last. You're not jumping into the deep end, you're gradually building confidence and skills. Most importantly, you're learning by doing, not by reading about what you should be doing.

The AI revolution isn't waiting for you to feel ready. But with these steps, you can catch up faster than you think.

I'd love to hear how you're learning AI. Drop me a line and let me know how it goes.

Honestly? The truth is the opposite.

Yes, I work long hours. And yes, entrepreneurship comes with tons of ambiguity, plenty of financial pressure, and way more uncertainty than I'd ideally choose. But here's what people don't realize:

1. I love my work. Most days, it genuinely doesn't feel like work. The post-bedtime shift? That's often my favorite part of the work day. My husband and I sit side-by-side on the couch, trash TV humming in the background, and we both just get shit done. It's strangely peaceful and deeply satisfying.
2. I have agency. This might be the biggest thing. I control my schedule, my travel, my location—everything. I work from home, see my kid during lunch and coffee breaks, take calls on dog walks, and shift my deep work towards afternoons/evenings when I'm most productive. That flexibility and control over my life make me better at both work and motherhood.
3. I'm showing my daughter I'm brave. One of my personal values (and one of our family values) is risk-taking. I believe deeply that better outcomes come when we take more risks. It's important to me that my daughter sees this, and I hope this career path helps her to.

Now, why am I telling you this?

I believe we're entering a moment, powered by AI, where this kind of career agency and fulfillment becomes accessible to more moms than ever before.

In the past two decades, we've seen online marketplaces like Etsy unlock millions of micro-businesses. Social platforms like Instagram and TikTok empowered influencers (including moms who now make a living sharing their lives). AI is the next massive shift, and it will be bigger than the ones before because it dramatically reduces the costs, complexity, and risk of being an entrepreneur.

For moms, the barriers to launching our own ventures have been steep: limited time, high startup costs, lack of technical expertise, and big opportunity costs. AI knocks down these barriers. According to a McKinsey report, AI could automate up to 60% of everyday tasks, freeing us to do more meaningful, strategic, and creative work.

How AI Is Already Changing the Game

I use AI to build my own AI-native startup, Lumi, and multiple micro-projects with zero technical training. The difference now is speed and ease:

• Creative ideas become actionable in hours, not months. I use Lovable to build websites with one prompt. (Short video explaining how here)
• Low costs mean less financial risk to experiment. Most AI tools are free or cheap, especially compared to hiring an engineer, designer, etc.
• AI automates repetitive work, preserving energy for high-impact tasks. If you haven't yet, ask ChatGPT how to save time in your day job to win hours for your side hustle.
• Better data on potential success. AI has made the internet much more accessible. You can now ask AI for data on which products in your space are selling the best, or what people are searching for on google to help guide your planning.

Data backs this up: Shopify found side hustles surged by 40% since AI became widely accessible, and women-owned businesses represent a significant portion of this growth.

Making It Real

I want to make this feel real and doable, so I'm sharing a few starter ideas for AI-powered businesses that would be relatively low-lift, high-fun, and (I think) very doable for a mom with a passion.

Right now, I'm bullish on Last Mile AI: companies that enable more AI creativity/fun by doing the last mile manual part that consumers don't want to do. Like...

• An AI Pet Portrait Service: AI can generate hilarious, share-worthy pet portraits, but the hassle of printing, framing, and shipping stops most people. I would spend silly money on framed 18th-century British aristocracy-style portraits of my dog, IF someone did the work for me.
• AI Baby Products: Generate unique designs or personalized books using AI prompts. Low upfront costs, easy to market on Instagram or Etsy, and loved by moms who value something special and unique. The number of hours I spend pondering unique baby gifts only to get thwarted by time to execute. Let me enter gender, age, name, and give me 3 cute ideas that you can execute for me.
• AI Travel Concierge: Trip planning can be so much more personalized and easily researched in the AI era. I'd pay for someone to create an AI itinerary and then generate a beautiful route map, personalized guide to the city, and maybe a post-trip personalized souvenir we could add to the family album or gallery wall.

Five Steps to Start Your AI-Powered Micro-Business

Have I inspired you yet? Well, let me share how easy it is to get started! Here are 5 things you can do right away:

1) Start dreaming

You need an idea you're genuinely excited about, so you have the will to stick with it long enough to reach profitability (which could be fast!).

Prompt ChatGPT:

"Help me brainstorm ways I could make enough money with a small business to quit my day job. Ask me questions to better understand what I'm good at and motivated to do. Include AI-enabled ideas."

2) Outline your plan

Take a few of your favorite ideas and use ChatGPT to outline what the first few months would look like. This helps you understand the reality of what it takes and what excites or scares you:

"Create a 90-day plan for how I could get started with [one idea]."

3) Understand the money

Use ChatGPT to make a financial plan for your idea. Have it research market pricing, competitors, benchmarks, and estimate your costs. Find out how many customers/products you'd need to break even and potential timelines:

"Research pricing, competitors, and benchmarks for [your idea]. Estimate costs, how many customers/products I'd need to sell, and potential timeline to profitability."

4) Let AI do the work

Now that you have a plan, ask ChatGPT (GPT o3 is best):

"Which of these tasks can you do for me? For tasks you can't do, explain how I can do them with affordable AI tools quickly."

Example: My favorite AI-powered website builder is Lovable. I'd say to ChatGPT: "I want to run with the xyz idea. I'm going to build the website in Lovable. Create plain English instructions I can paste into the Lovable prompt box to generate the website." Then literally just copy-paste into Lovable (it's free to start) and watch the website come to life.

5) Start marketing immediately

Use AI to create a marketing plan and content quickly and cheaply. It helps test demand, build awareness, and gain momentum at zero cost:

"Create a simple marketing plan and five pieces of content to test demand for my business. Tell me where to post them and when to generate the most views."

Now's Your Moment

AI has lowered the bar dramatically—less financial risk, less complexity, less overwhelm. Most importantly, more room for creativity, joy, and freedom.

You don't have to risk it all or quit your job immediately. You just have to start playing. The rewards—financial, emotional, professional—can be game-changing. As a mom, it's not just an opportunity; it feels like the moment we've been waiting for.

Ping me with your ideas, tell me how I can help—I'm here to make this real for more moms.

Let's make it ours.

As a non-technical founder in 2025, I'm experiencing a transformation that's hard to overstate. In the past, I was dependent on engineers to bring my vision to life. Now I'm building complex features independently, sometimes in a matter of hours.

This isn't just about building faster—it's about fundamentally changing what's possible for non-technical founders. Here's my experience living on this frontier.

Direct from Customer Call to Working Feature

My product development process has completely transformed. I used to sketch ideas, create detailed specs, hand them to engineers, wait, give feedback, wait some more, and then finally see my vision come to life—usually different from what I imagined.

Now, I go straight from customer insights to building. Here's my current flow:

• After a customer call reveals a new insight, I immediately open Lovable
• I start building as a way to think through the idea
• I typically get to a working feature within hours, or I realize quickly it's not the right thing and move on
• I ship it to users
• I iterate rapidly based on their response

This hands-on approach means I'm no longer translating my vision for others—I'm directly manifesting it. The Win Tracker went from idea to live product in 2 days. For non-technical founders, this is like suddenly growing wings after years of asking for rides.

My Founder AI Tech Stack

• Lovable: No-code AI development tool
• ChatGPT: Personal AI advisor and content generator
• Granola: Meeting recording and transcription tool
• Lumi: Tracking my wins and coaching me
• Gamma: Occasional slide deck assistance
• AI PR strategist (a.k.a. "Chelsea"): Custom PR expert built with ChatGPT

The Perfect Imperfection of AI Development

There's a beautiful alignment between early-stage startups and AI development tools: neither needs to be perfect to be valuable.

Conor, my part-time engineer, often points out technical issues in what Lovable produces, and he's right. But from my founder perspective, those issues usually don't matter yet. I'm focused on validating whether users even want the feature, not whether the code is optimized for scale.

This is perhaps the most liberating aspect of building with AI. I can create "good enough" solutions that let us test concepts with real users. Later, once we know what's worth investing in, we can rebuild with engineering rigor.

In the early days of a startup, speed of learning trumps perfection of execution. AI development tools are perfectly imperfect for this phase.

How I Build with Lovable

The magic of Lovable is that it functions similarly to ChatGPT—you simply tell it what you want to build. No wireframing, no mockups, no complex specifications. Just type in natural language what you need, and it generates the code and interface for you.

For example, I'll type something like: "Create a page that shows a user's wins based on meeting transcripts" or "Add a button that sends an email notification when a new win is logged." When I want to make changes, I just describe them: "Make the button blue instead of green" or "Add a section for superpowers beneath the wins."

What's particularly powerful is how Lovable walks you through connecting to the "backend"—the database to store user data, or integrations with other tools like Resend for automated emails. Instead of learning SQL or API documentation, I simply follow Lovable's conversational prompts to set up these connections.

So far, I've built:

• A homepage with a waitlist signup form that automatically tracks signups and sends confirmation emails
• Event pages with automations for registrations and sending Zoom invites (I just tell Lovable which new events to add)
• This blog you're reading right now
• The Win Tracker and other AI-powered career tools that analyze meeting transcripts and provide insights
• Our AI coach chat interface that gives personalized career guidance

As we test, iterate, and eventually scale, we'll likely move away from Lovable for some components. But for now, I continue to be surprised by how far I can push this no-code approach to product development. Each week, I test its boundaries and discover I can build something I previously thought required an engineer.

Expanding Beyond Product Development

My newfound technical agency extends beyond product development. I created an AI PR strategist I call "Chelsea" simply by prompting ChatGPT to take on that role.

I told it: "You are Chelsea, my PR person and expert in early-stage consumer AI startup PR. Your goal is to help me get loads of users flowing into our product. Come up with a full plan and specific action items, telling me which you can handle and which I need to do."

Chelsea responded with a comprehensive PR strategy and the assets I needed to execute it. While I still need to implement much of the plan myself, I now have expert-level strategic guidance at my fingertips, without hiring a PR firm.

This pattern repeats across business functions—marketing, strategy, operations. The bottleneck is no longer access to expertise, but time to execute.

A New Era for Non-Technical Founders

For anyone considering entrepreneurship without a technical background, my message is simple: there has never been a better time to start.

You no longer need to:

• Find a technical co-founder before validating your idea
• Raise money to hire engineers for an MVP
• Learn to code yourself (please don't do this, huge waste of time)

What you do need:

• Clarity about the problem you're solving
• A willingness to dive into AI tools like Lovable
• Scrappiness and a "get started" mentality

The democratization of building means the primary constraint is now your vision and understanding of user needs, not your technical ability.

I still value developers immensely—there are absolutely things AI can't do well, and their expertise is crucial for scaling and stability. But I'm no longer blocked waiting for technical resources to test ideas and iterate.

In 2025, non-technical founders can go from zero to one largely by themselves. And that changes everything about who can build the future.

"Two years ago, I had to manually log every meal and got generic advice," she told me. "Now I snap photos, and my day just…improves."

We're entering an era where the Return on Data Invested (RODI) is extraordinarily high, especially for high-achieving professionals who thrive on efficiency and results.

Data Has Shifted From Exhausting to Empowering

In the past, personal data often felt exhausting. We logged calories, tracked steps, and recorded meetings with minimal reward. Today, AI has completely changed the equation:

• Effortless Capture: Data now flows passively into apps and platforms, requiring minimal effort from you.
• Deeper Understanding: AI identifies patterns and insights you'd likely never discover on your own.
• Automatic Action: Crucially, AI doesn't stop at insights—it proactively takes action on your behalf.

Together, these three factors turn personal and professional data into powerful, actionable assets.

From insights to action: real examples of RODI

High-achieving professionals across industries are already quietly leveraging their data to maximize outcomes:

• Health: Wearables no longer just track sleep—they adjust your bedtime routine automatically, ensuring you wake energized and ready for peak performance.
• Finance: Personal finance AI doesn't just categorize spending—it proactively negotiates lower bills and better deals on subscriptions.
• Productivity: Meeting transcripts that automatically generate follow-up emails and project summaries, cutting tedious admin tasks from hours to minutes.

Each example involves minimal ongoing effort, yet produces significant, compounding returns.

Professionals are already pushing the boundaries—because it's worth it

Today, many ambitious professionals are entering ethical grey areas, like recording meetings or capturing sensitive conversations. Why? Because the career advantage they gain from actionable insights and automation dwarfs potential risks.

One VP recently admitted to me: "In six months, everyone will be doing this. How could you not, when the upside is massive?"

Practical ways high achievers leverage their career data today

Consider these scenarios—all based on data you likely already generate every day:

• Influence and communication: By analyzing your Slack and email history, AI agents identify exactly how to frame your ideas to resonate with senior stakeholders.
• Networking and relationships: Calendar and email data pinpoint which professional relationships are most critical and automate personalized touchpoints, ensuring you maintain strong connections effortlessly.
• Energy and productivity: Your meeting and activity data uncover your productivity peaks and valleys, enabling proactive calendar adjustments to match your natural rhythms.

For example, one professional discovered through AI analysis of her meeting recordings that executives consistently tuned out detailed presentations after three minutes. With this insight, she began leading with high-level takeaways, immediately improving engagement and approval rates.

A glimpse into what Lumi is building

At Lumi, we see this trend clearly. We're developing tools that turn your everyday professional data—emails, transcripts, calendars—into actionable insights and automated assistance. For instance, Lumi analyzes your meeting patterns and communication style to automatically create concise briefings or follow-up notes tailored to your stakeholders' preferences.

The goal isn't just to inform you, but to directly improve your outcomes, without adding work.

Audit Your Data Vault This Week

The biggest career step you can take in 2025 might simply be auditing your own "data vault":

• What professional information are you already generating that's currently idle?
• Could analyzing your meeting transcripts, email patterns, or calendar data unlock insights you can immediately act upon?
• Which tedious career tasks—networking, follow-ups, project tracking—could your data start managing for you?

This isn't the future. It's happening now. Your data is finally more valuable than the effort it takes to share it. It's time to let it work for you.

Let's cut to the chase: the middle of the professional bell curve is exactly where AI will hit hardest. But this isn't the career apocalypse many fear. It's the greatest opportunity for talented professionals to escape the corporate hamster wheel and build something uniquely theirs.

I'm calling this shift "The Spike Factor" and it's going to completely rewrite the rules of knowledge work.

The Great Professional Reshuffle

For decades, organizations have pushed us toward standardization. The same MBA frameworks, the same consulting methodologies, the same corporate playbooks. The safe bet was becoming well-rounded and competent across a predictable set of business skills.

AI is flipping that equation:

The conventionally successful corporate generalists who are "pretty good at PowerPoint," "solid at Excel," and "effective communicators" are now competing with algorithms that can execute those standard business functions with greater speed, consistency, and scale.

Meanwhile, the professionals with distinctive, impossible-to-replicate combinations of expertise, perspectives, and specialized knowledge are becoming exponentially more valuable.

This isn't theoretical. Just look at what's happening to entry and mid-level jobs in finance, consulting, marketing, and legal. Tasks that once occupied junior associates' first two years are increasingly automated. The standardized parts of professional work are vanishing before our eyes.

Your Professional Edge is Your Economic Engine

Your spike isn't just a specialization. It's your unique professional fingerprint. It's where your particular combination of industry knowledge, technical skills, relationship networks, and lived experiences create a perspective no one else has.

Think about it like this:

The social media manager who creates standard corporate posts → fighting for survival against AI

The creator who combines Korean skincare expertise with data visualization skills to create hyper-personalized skincare analytics for beauty influencers → booked and busy

The difference isn't about credentials or even experience. It's about the uniqueness of your professional angle. The narrower and more distinctive, the more valuable.

The Etsy Economy on Steroids

Remember what Etsy did? It turned millions of crafters with specific, often niche obsessions into entrepreneurs. People making hand-carved wooden spoons or custom dog portraits suddenly had global markets for their weird, specific talents.

What's coming is Etsy for knowledge workers.

AI is demolishing the overhead costs of running a professional service business. Tasks that once required a support staff (research, document preparation, logistics, basic client management) are increasingly handled by one person with the right AI tools.

This means we're about to see an explosion of micro-consultancies and specialized professional service providers built around hyper-specific knowledge:

• The former consultant who combines supply chain expertise with sustainability metrics to become the go-to advisor for fashion brands transitioning to ethical manufacturing
• The ex-pharma marketing executive who leverages her regulatory knowledge to create specialized AI compliance frameworks for healthcare startups
• The corporate attorney who becomes the definitive legal translator between Japanese business culture and Western tech companies expanding into Asia

These are exactly the kinds of hyper-specialized services that will thrive in the next decade.

The Free Time Economy

Here's the other massive shift nobody in the C-suite is talking about: AI isn't just changing what work we do, it's dramatically reducing how much we need to work. Analysis that used to take weeks now takes hours. Reports that took days can be compiled in minutes.

This creates two unprecedented opportunities for professionals with the right spike:

First, you can deliver more value in less time, making solo operations far more profitable than traditional business models. The margins on expertise are about to get ridiculous when you don't need a team to operationalize your insights.

Second, and this is where things get really interesting, people will have more discretionary time than we've seen since before the industrial revolution. More time means more interest in learning, development, exclusive experiences, and self-actualization. The affluent professional class will pay premium prices for experiences that help them maximize their newfound freedom.

Think about the possibilities:

• The former management consultant who creates executive retreats combining business strategy with wilderness survival skills.
• The finance professional who designs personalized financial literacy programs for creative professionals.
• The ex-HR executive who builds transition programs for professionals looking to escape corporate life and find more meaningful work.

The future of work isn't just less work. It's more meaningful connection, more personal growth, more of what makes us distinctly human. And that creates endless opportunities for knowledge workers with the right spike.

Finding Your Spike: A Practical Guide

So how do you actually find your spike? Here's my practical approach:

• The Compliment Audit: What do people consistently praise you for that feels effortless to you but seems to impress others? Write down the exact phrases people use. Or get lumi and we'll track this for you automagically.
• The Obsession Inventory: What rabbit holes do you fall down that others find boring or bizarre? What topics make your friends' eyes glaze over but get you fired up? These aren't distractions, they're clues.
• The Frustration Flip: What drives you crazy about your industry that nobody else seems bothered by? Your irritations often reveal your values and unique perspective.
• The Skill Stack Analysis: List your top 5-7 skills (not just professional ones). Now look for unusual combinations. The magic rarely lives in any single skill, it's in the overlap of several.
• The Market Test: Start small. Create something (a blog post, a service, a prototype) that showcases your unique angle and see who responds. The market will tell you if you're onto something.
• The Conversation Test: If you talk about this potential spike at a party, do people ask follow-up questions or do they change the subject? Genuine curiosity from others is a good sign.
• The Energy Check: After spending three hours working on this potential spike, do you feel drained or energized? Sustainable spikes give you energy rather than depleting it.

Living in the Professional Gaps

The most valuable opportunities aren't in crowded, established professional lanes. They're in the unusual spaces between them. The investment banker who moonlights as a fiction writer and helps creative studios structure financing deals. The marketing executive who leverages her passion for architectural history to specialize in heritage property branding. The operations leader whose weekend cycling obsession led to a breakthrough methodology for supply chain optimization for sporting goods companies.

These intersections are where the highest-value work happens because they're harder for AI to reach and harder for competitors to replicate. AI excels at optimization within established domains. Humans excel at connecting seemingly unrelated dots and developing novel frameworks.

The Bottom Line

The conventional professional playbook (MBA, prestigious firm, corporate ladder) isn't dead, but it's no longer the safe bet it once was.

The future belongs to the specialists, the polymaths, the professionals who embrace their distinctive edges rather than smoothing them out to fit corporate expectations.

The question isn't whether AI will take your job. The question is: are you ready to leverage AI to build a more profitable, impactful, and personally fulfilling professional life around what makes you uniquely valuable?

Your professional distinctiveness isn't a liability anymore. It's your greatest economic asset. Start investing in it now.

What's your professional spike? What unusual combination of industry knowledge, technical skills, and lived experience might be your economic superpower? Drop it in the comments. I bet it's more valuable than you think.

This post is part of my ongoing Work Waves series examining how AI is reshaping the professional landscape.

The AI revolution will unfold in three distinct waves – and the professionals and companies who understand this pattern will be positioned to thrive.

Wave One: The End of Career Friction (2025-2026)

The first wave is already crashing onto our shores. Generative AI is eliminating the friction from professional advancement virtually overnight.

Think about the career tasks that professionals universally dread:

• Updating resumes and LinkedIn profiles
• Crafting networking outreach messages
• Preparing for interviews
• Asking for raises and promotions
• Building professional visibility

These "career chores" have been persistent pain points – tedious, anxiety-inducing activities that most professionals either avoid entirely or complete with minimal effort. The psychological barriers are enormous, and the opportunity cost of doing them poorly is even larger.

AI is removing these barriers completely. Tools already in market can instantly generate personalized networking messages, create tailored resumes for specific roles, prepare customized interview responses, and craft compelling professional narratives that once took hours of agonizing work.

But the truly revolutionary change isn't just automation – it's the psychological transformation. When career development becomes effortless, intuitive, and even enjoyable, professionals develop entirely new relationships with their career trajectories. The mental energy previously spent on career anxiety gets redirected to strategic thinking and meaningful advancement.

I've seen this shift firsthand in our early testing. Users describe it as "career superpowers" – the feeling that they're suddenly operating at an entirely different level than their peers. One early tester put it bluntly: "This makes me feel like I have a professional hype team and expert agent working for me 24/7."

This first wave alone would represent a massive shift. But it's merely the prelude.

Wave Two: The Great Career Reimagination (2026-2030)

The second wave is where the deeper transformation occurs. AI won't just change how we manage our existing careers – it will force millions to completely reimagine what they do for work.

McKinsey projects that 12 million U.S. workers will need to change occupations by 2030 due to AI-driven disruption. Goldman Sachs estimates roughly 300 million jobs worldwide could be affected. The World Economic Forum predicts 83 million jobs eliminated and 69 million new ones created by 2027.

These aren't just statistics – they represent an unprecedented scale of career upheaval:

• Knowledge workers watching core job functions being automated
• Mid-career professionals needing to pivot to adjacent or entirely new fields
• New roles emerging that combine human skill sets with AI capabilities
• The rise of entirely new career categories we can barely imagine today

My hypothesis is that we'll see a surge in solo entrepreneurship during this period. As companies accomplish more with fewer employees, individuals empowered by AI tools will discover unique ways to create value. The combination of automation handling routine tasks and AI augmenting human creativity will enable many to develop their "unique thing" and make a living from it.

Think of it as Etsy and Shopify on steroids, but with a stronger service component. We'll see designers, consultants, creators, and specialists of all kinds building micro-businesses around their distinctive talents. Women, in particular, will gravitate toward this model because it provides the autonomy and agency over schedules that traditional employment often restricts. Entirely new platforms, products, and communities will emerge to support this distributed workforce – creating infrastructure for a fundamentally different relationship between talent and the market.

According to PwC, 44% of workers' core skills will need updating by 2027. We're talking about a fundamental reinvention of professional identity for nearly half the workforce.

Many see this as a crisis. I see it as perhaps the greatest opportunity in a generation for professionals to align their work with their interests, values, and strengths.

Wave Three: The Post-Work Society (2030+)

As we move toward more advanced artificial general intelligence (AGI), we'll enter an even more profound transformation. While Waves One and Two change how we work, Wave Three will challenge why we work at all.

In this third wave, AI will have automated not just routine tasks but entire complex workflows. The traditional 40-hour workweek – a relatively recent invention in human history – will become increasingly unnecessary for economic productivity. We'll face the existential question of how to structure our lives when most economic production can be accomplished with minimal human labor. The challenge won't be finding work, but finding purpose and meaning in a world where traditional careers no longer serve as the primary organizing principle of adult life.

The Critical Connection Between Waves

Here's the key insight that's often missed: success in Wave Two will be determined largely by who masters Wave One.

The professionals who adopt AI career tools early will develop key advantages:

• Continuous adaptation mindset: They'll build the habit of evolving with technology
• Career agility: They'll have systems in place to identify opportunities faster
• Personal brand clarity: They'll have well-defined, adaptable professional narratives
• Network leverage: They'll maintain stronger professional connections that surface opportunities
• Skills visibility: They'll have clearer insights into which skills transfer to emerging fields

Early adopters of AI career management won't just have better tools – they'll have fundamentally different psychology around professional change. They'll approach career pivots with confidence rather than fear, with strategic clarity rather than paralysis.

What This Means for Professionals, Companies, and Investors

For individual professionals, the message is clear: the time to embrace AI career tools is now, not when your role is already being disrupted. The early adopters will have tremendous advantages in navigating the coming transition.

For companies, this transformation presents both challenges and opportunities. Organizations that help employees adapt will maintain incredible advantage in talent retention and institutional knowledge. Those that ignore the coming wave will face painful transitions and talent losses.

For investors, the career space represents one of the most compelling opportunities in the AI landscape. The market for AI-powered career development is potentially enormous – hundreds of millions of professionals worldwide facing unprecedented career change, seeking tools that give them edge and clarity.

Leading Through the Transition

I'm dedicating the next chapter of my career to building at this critical intersection – creating the tools that will help professionals not just survive but thrive in this new landscape.

The foundations are in place: the technology has matured, professionals are seeking solutions, and the market is ready for category-defining products that turn career anxiety into career advantage.

The AI career revolution isn't coming – it's already here. And I couldn't be more excited to help shape how we navigate it together.