It worked fairly well.
This page is the practical follow-up: how to decide what you want from time away, what AI should handle, and how to build a system that can change with you.
The reality of time away
A couch, a stroller, and two ways to use technology.
The thing about maternity leave is that, at least in my experience, you spend a ton of time sitting on the couch holding a baby. And eventually, a lot of time taking the baby on walks.
When it comes to technology, that gives you two main modes: one-handed phone scrolling from the couch and voice-only mode on walks. And you have a lot of time in each of them.
To be clear, I spent 90% of my couch time watching Love Island and 90% of my walk time listening to beach reads. But for that 10% when I was motivated to work, I had never felt more productive and capable.
So here are my recommendations if you’re thinking about your own time away from the desk.
First, and most importantly
Define your goals
A system designed to preserve revenue looks very different from one designed to protect complete time off. You might want the business maintained, useful work waiting for your return, or a little room to move selected ideas forward. Clear goals stop AI from creating busywork simply because it can.
Questions to clarify what you want
- What must remain healthy while I am away?
- What can pause without real consequence?
- What would I be grateful to return to?
- How available do I genuinely want to be?
- Which decisions or activities do I still enjoy and want to keep?
- What would make this time feel successful personally and professionally?
Write down a short set of goals and constraints before you design any technology.
Second
Decide who should do what
Sort the work that could support those goals into three groups. This is honestly the bulk of the work. You do not need perfect answers; you need a starting hypothesis that the first test can improve.
- AI can handle it: Recurring, bounded work with clear inputs, outputs, and approval limits.
- AI and I do it together: Exploration, judgment, or work that becomes better through conversation and review.
- I want to keep it: Decisions, relationships, creative work, or anything I value doing myself.
Use ChatGPT’s best available model with its highest reasoning setting. This planning benefits from more time to think.
Turn my goals into a plan
Help me design a lightweight system for time away from my desk. Start with my goals, then decide what AI can handle, what we should do together, and what I should keep.
Read the full planning prompt
Help me design a lightweight system for time away from my desk. Start with my goals, then decide what AI can handle, what we should do together, and what I should keep.
START WITH WHAT YOU KNOW
Use what you already know about me and my work, plus documents, conversations, and sources I have shared or connected. Show me your stated assumptions, tell me what you could not determine, and ask me to fill only the important gaps.
UNDERSTAND MY GOALS
Help me clarify:
- What must remain healthy while I am away.
- What can pause without real consequence.
- What I would be grateful to return to.
- How available I genuinely want to be.
- Which decisions, relationships, and activities I want to keep.
- What would make this time successful personally and professionally.
- My available time, budget, tools, comfort with setup, and preferred ways of working from a phone, by voice, or at a computer.
DIVIDE THE WORK
Sort relevant work into three groups:
1. AI can handle it: recurring, bounded work with clear inputs, output, and approval limits.
2. AI and I do it together: exploration, judgment, or work improved by conversation and review.
3. I want to keep it: decisions, relationships, creative work, or activities I value doing personally.
Explain your reasoning and flag anything that is risky, ambiguous, or dependent on information you cannot access.
CHECK WHAT IS POSSIBLE NOW
State today’s date. If you can browse, check current official product documentation before recommending tools. Link to the sources and tell me when you checked them. Verify the features that matter, including phone and voice access, connected sources, scheduled tasks, work that can continue with the app closed, email delivery, and cost. If you cannot browse, say so and ask me for current documentation rather than presenting old knowledge as current.
BUILD THE SMALLEST USEFUL PLAN
Recommend the simplest system that supports my goals. Do not recreate a complex agent system unless the work requires it. For each recommended workflow, specify:
- The result and why it matters.
- The context and sources it needs.
- One suitable tool and a simpler fallback.
- Setup effort and likely cost.
- What AI may do and where it must pause.
- What I need to review.
- A small first test and criteria to keep, change, or stop it.
BOUNDARIES
Sending messages, spending money, publishing, contacting people, launching offers, and making commitments require my explicit approval. Do not claim a workflow is active until it has been set up and tested.
Return a prioritized implementation plan. Begin by showing me what you think you know, your assumptions, and the gaps you need me to fill.
Third
Design the system to change
Models improve, new tools arrive, and your priorities change. This part is hard. I recommend a monthly routine where AI reviews the latest developments and tells you where the system could meaningfully improve, without changing anything on its own.
Set up a monthly system review
Once a month, review this system against my current goals and official product updates. Email me only the changes worth considering and the decisions you need from me.
Read the full monthly review prompt
Create a monthly review routine for my AI system. Its job is to tell me what should change, not to make changes without my approval.
REVIEW THE CURRENT SYSTEM
Once a month, revisit my goals, constraints, priorities, workflows, tools, costs, and decision log. Identify anything that has stopped working, become unnecessary, or no longer supports what I want from my time away.
CHECK CURRENT DEVELOPMENTS
State today’s date. Review official product updates and documentation for every tool used in the system. Link to your sources and record when you checked them. Identify relevant model improvements, new capabilities, removed features, price changes, privacy or permission changes, and simpler supported approaches. Do not rely on old knowledge or general AI news when official information is available.
MAKE SELECTIVE RECOMMENDATIONS
Recommend a change only when it creates a meaningful benefit. For each recommendation include:
- What changed and the official evidence.
- Which workflow it affects.
- The practical benefit.
- Migration effort, cost, and risk.
- Assumptions or information you could not verify.
- Whether to change now, test first, or do nothing.
- The decision you need from me.
DELIVER BY EMAIL
Send me one concise email once a month with four sections: what changed, what it affects, your recommended action, and the decisions you need from me. If nothing meaningful changed, say so. If the tool we are using cannot schedule this routine or send email, tell me and help me choose a currently supported way to do it. Do not pretend the routine is active until we have tested delivery.
BOUNDARIES
Do not replace tools, edit workflows, spend money, publish, contact people, or change permissions without my explicit approval. Keep recommendations separate from authorized actions.
Finally
Practice voice mode
I’m serious. Voice mode is a behavior change, which means it feels weird when you start. Practice before your leave instead of expecting it to feel natural the first time you need it.
Try thinking through an idea on a walk, dictating the context AI needs, pushing back on a recommendation, and ending by asking for a written summary or next step. Once it clicks, you’ll love it for everything. Voice is what makes the system usable when a keyboard is not.
A few more ideas
Two of my favorite agents from Maternity Mode
Hopefully the recommendations above give you a personalized system for your leave. If you need inspiration, these were two of the most useful ways I put mine to work.
The Opportunity Scout
One of my favorite agents was an Opportunity Scout. It investigated new business opportunities that fit me and brought me one developed recommendation each week.
It brought me genuinely relevant, strong opportunities that I was excited about and wouldn’t have found on my own.
The important part was not the technical setup. The Scout had enough context about my products, assets, goals, constraints, and past decisions to make a useful recommendation. It researched what it could access, showed its evidence and assumptions, and sent me a weekly email with the decision it needed from me.
It looked for the overlap between real demand, my strengths and existing assets, and something that would fit my leave: asynchronous delivery, little maintenance, and no live sessions.
Set up your own Opportunity Scout
Help me set up an Opportunity Scout for my business. Its job is to bring me one developed recommendation each week.
Read the full Scout prompt
Help me set up an Opportunity Scout for my business. Its job is to bring me one developed recommendation each week that helps me decide whether an opportunity is worth testing.
START WITH WHAT YOU KNOW
Use what you already know about me and my work, plus any documents, conversations, or sources I have shared or connected. Draft the business context first. Show me your assumptions, tell me what you could not determine, and ask me to fill only the important gaps.
UNDERSTAND THE BUSINESS
Build a concise working brief covering:
- Who I serve and the problems they come to me with.
- What I offer and already have, including services, products, content, skills, relationships, and reusable assets.
- What I am trying to achieve now.
- My constraints, including time, budget, delivery model, and the maintenance I am willing to take on.
- Ideas, commitments, or offers I have ruled out.
FIND USEFUL OPPORTUNITIES
Look for the overlap between real demand, my strengths and existing assets, and something practical under my constraints. Use relevant sources I have provided or approved, plus public sources you can cite. Separate facts from inferences and assumptions, and state what you could not access.
SEND ONE WEEKLY BRIEF
Send this to me as an email once a week. Include:
- The opportunity, buyer, and problem.
- Why it fits me and my business.
- The evidence, gaps, and assumptions.
- The effort to build and maintain it.
- The cheapest useful test.
- The decision you need from me.
If nothing is worth testing, say so. Do not repeat rejected ideas unless meaningful new evidence changes the case. If you cannot schedule or send email from the tool we are using, tell me and help me choose a currently supported way to do it.
BOUNDARIES
You may research and recommend. Ask before spending, launching, publishing, contacting people, or making commitments. Do not act on a recommendation until I approve it.
Prepare for my return
While I was on leave, I had lots of ideas for the business. Things like: I should be on more podcasts. Or: I should focus my consulting work on coaching and leadership-development companies. Any time I had a big idea like this, I felt the urge to act on it.
Instead, I gave the idea to ChatGPT or GrokBot in a simple one-shot prompt. For podcasts, I’d say: “I want to be on more podcasts. Make a list of relevant podcasts I should pitch, including names and why each one is a fit. Make it all ready for me to action when I return.”
I did the same thing for coaching and leadership-development companies. I had AI build a list of 100 firms with useful details, a “why them,” and a “why me.” It also identified possible warm connections and prepared draft emails for me to review. When I returned, the research was organized and ready. If I wanted to move forward, I could make the decisions and hit send.