Join Mainframe Mom

At the bottom: one question to help you find the win hiding inside your own numbers.

On August 31, I opened Stripe and Etsy to gather the numbers for my month-end post.

Three numbers were staring back at me:

• 72 Instagram followers

• 13 email subscribers

• $285 in August sales

No giant launch. No high-ticket offer. Just subscriptions and low-ticket products.

No, $285 is not life-changing money.

But it changed something for me.

A small audience is not a failure when you’re beginning to find the right people.

Small does not mean failed

I’ve made my goal public on my Building in Public scoreboard: build three products, sell them, and earn Mainframe Mom’s first $1,000 by early October.

This $285 belongs on that scoreboard.

To be clear, I can’t draw a clean line from my 72 Instagram followers to every dollar I made. The sales came through Stripe and Etsy, so this is not a case study claiming that 72 Instagram followers generated $285.

What it does prove is that real people are willing to pay for things I created.

That matters because I didn’t make these products just to have something to sell. I created them for myself first. They are systems I use in my own house. Systems I was proud of and believed in enough to share with other moms.

Then other people found them and decided they could help them too.

That is the proof I needed.

A following is not the same as a business

It’s easy to believe success starts after you reach a certain number.

After 1,000 followers.
After 10,000 followers.
After your email list looks less like a group chat and more like an actual audience.

I’m building this brand in public from scratch, and growth has felt slow. On the exact same night that Stripe and Etsy showed me $285, Instagram still showed 72 followers and my email list still had only 13 names on it.

The audience had not suddenly become impressive.

But the products had started finding people who wanted them.

That doesn’t mean audience size is irrelevant. I need to reach more people if I want this business to grow. It means a small number is not automatically evidence that what I’m building is failing.

Sometimes, the earliest proof is simply that a few people cared enough to act.

Count the people, not just the number

Maybe you gained only 10 followers this month.

It’s easy to see “10” and think you’re barely moving.

But those are 10 individual people who stopped, listened, and decided they wanted to hear more from you.

Ten people are not a viral audience but picture 10 actual women standing in your kitchen because they believe what you’re building might help them.

That feels different.

The $285 did not transform my finances. It gave me something I needed just as much at this stage: motivation to keep showing up, keep creating, and keep finding more of the people who need what I’m making.

Every follower, subscriber, and customer is a real person choosing to come along with you.

And every single one of them is a reason to keep going.

Small does not mean failed. Small can mean you’ve found your first proof.

What is one small win you almost dismissed this month?

Leave it in the comments. I want to celebrate it with you.

Leave a comment

If you’re building something of your own alongside everything else you carry, subscribe to Mainframe Mom. You can grow slowly here.

Get the next sharp take

Paid members get the systems, templates, and behind-the-scenes builds, not just the ideas.

If you onboarded an employee and they got 70% of a complicated job right on their first try, you would think they had done a pretty damn good job.

If a person missed the other 30%, you would call it training.

When AI misses it, most people call the whole thing a failure and take the job back.

I know because every single household agent I have built started at about 70%.

My meal planning agent kept giving me the same five recipes every week. My family was not going to eat that. My bookkeeping agent could not separate CSV files.

Neither one was ready on the first try. Neither one was useless, either.

So in this paid build, I am giving you five small, low-risk household handoffs you can make today, plus the exact prompt to start each one. Because the fastest way to get household work out of your head is not to begin with the biggest job.

The reality is that 70% is still pretty darn good. It is not a failed handoff. It is a first day of onboarding.

We keep starting at the top

Most moms begin with something massive:

Become my meal-planning agent.

Organize my household finances.

Then AI returns something that is useful but not quite right. The meal plan is repetitive. The bookkeeping file is formatted incorrectly. Because the result does not create immediate, complete relief, we abandon it.

We are starting at the top of the ladder when we should be starting at the bottom with the small annoyances that repeat often enough to take up permanent space in our heads.

I now rank possible handoffs with two questions:

1. How often does this happen? Weekly or more moves it up the list.

2. How much does it nag me? I score it from 1 to 3.

The things that happen at least weekly and score a 3 in the nag category go first. Then I choose the smallest piece of that work I can safely hand off.

The annoying 30% becomes the onboarding manual

Meal planning was one of my biggest annoyances, and it happened every week, so it was worth working through—even though it was a large first handoff.

“Plan my family’s meals” sounds like one task.

It is actually dozens of decisions wearing a trench coat.

What will everyone eat? What did we have last week? Which dinners fit the time we have? How much variety does my family need? Which vegetables do my kids stand a chance of touching?

My agent could not know those answers until I taught it.

So when it gave me the same handful of recipes again, I talked to it the way I would talk to a normal person who had completed the assignment but did not yet understand how my household works.

I told it:

• Give me a more diverse menu.

• Pull from at least 15 different meals.

• Break up similar meals as much as possible.

• Never give me the same meat two days in a row.

• Never choose anything that takes longer than 40 minutes to cook.

• Aim for about two Crock-Pot meals each week.

• Aim for at least two slab-of-meat-and-starch meals each week.

• Include vegetables my kids would actually eat.

The annoying 30% became the household onboarding manual.

The next meal plan was exactly the way I wanted it.

Now, every single week, my agent automatically drafts our meal plan and tells me when it is ready. I approve it. Then it creates the grocery list and writes a recipe for every meal.

The approval belongs to me. The recurring mental labor does not.

And no, AI does not need every private detail about my family to do this. I use first names and ages. I never provide last names, full birth dates or addresses. I do not even use my child’s pediatrician’s name; I call him Dr. B.

Keep sensitive information to yourself. Hand off the safe, repetitive work.

Here are five places to begin.

Read more

One practical AI system for ambitious moms, every Monday.

Join Mainframe Mom

The exact AI input is at the end.

Why settle for less when all it takes is two more minutes?

I asked AI to create a complete workflow for launching and monetizing a YouTube channel. The first version had 19 steps. It was substantial, thoughtful, and very close to what I wanted.

I could not see anything obviously missing.

I asked anyway.

What haven’t we considered that would help me start a YouTube channel from scratch and monetize it to its maximum potential?

AI surfaced roughly 30 additional considerations I had not thought to include.

After every important AI output, ask AI to critique and improve its own work against your actual goal.

The second pass found what I couldn’t see

The original workflow focused heavily on producing individual videos. The critique revealed that it was not focused enough on building a YouTube business from zero.

AI recommended adding a channel-level strategy before researching individual topics. It added a revenue hypothesis to every video brief, a launch phase specifically for the first 20 videos, a stronger packaging process, and revenue per 1,000 views as a performance metric instead of looking only at RPM.

It also suggested adding a post-mortem after every published video:

What worked?
What failed?
What will we change in the next video?

The first workflow was strong. The second pass made it more complete.

If you are not sure whether you need a quick answer or a repeatable system, I use this five-second test to decide when the work should stop living in chat and become a workflow.

Stop using AI like a vending machine

The plug-and-chug approach keeps people stuck using AI as a chat box.

Prompt goes in. Answer comes out. Conversation over.

A learning loop adds another step. AI evaluates the output against your stated goal, identifies possible weaknesses, and proposes a better version.

That does not mean the model permanently learns from one follow-up question. It is critiquing the current output inside the current conversation.

If you want the next run of a recurring workflow to improve, take the changes you approve and add them to the workflow’s instructions or evaluation checklist. That gives the workflow something concrete to use the next time it runs.

The loop becomes:

1. Create the output.

2. Critique it against the goal.

3. Approve the useful improvements.

4. Add those lessons to the instructions for next time.

That is how a one-time critique becomes a workflow that improves.

But maybe it is already good enough

Sometimes it is.

Not every grocery list needs a post-mortem. If the output is low-stakes, disposable, and already useful, take the win. The goal is not to manufacture extra work or chase perfection.

But when the output matters, or when the workflow will run repeatedly, two minutes of critique now can prevent the same corrections from landing back in your lap later.

The next time AI creates something for you, ask:

How can we improve this to help me achieve my goal? My goal is ______.

The first AI result can be good and still not be finished.
Two more minutes can turn it into a learning loop.

What is one AI workflow you want to improve before it runs again?

Get the next sharp take

Leave a comment

Join Mainframe Mom

At the bottom of this post, I’ll show you the first household task I would hand to AI if your brain is running the entire family.

One Saturday, my dad and stepmom took my kids to the trampoline park.

This does not happen often.

I had been given a free afternoon. I spent it proving I deserved one.

I had no work to finish. I had no children in the house. For once, nobody needed a snack, a ride, an answer, a clean shirt, or help finding the thing directly in front of them.

I thought I should sit down.

Then I saw the dishes in the sink.

I decided I could relax as soon as the sink was empty. The dishes led me to the laundry. The laundry led me to the living-room floor. The living-room floor became the entire house.

By the time I was finally ready to sit down, my kids were home and I was making their lunch.

A free afternoon is an unused muscle. When every minute has belonged to somebody else for years, you can forget how to use time that belongs only to you.

The to-do list makes rest feel irresponsible

For more than seven years, nearly every waking minute of my life has been accounted for by somebody else.

Honestly, many of my sleeping minutes have been accounted for too. My kids still wake up at night.

I need to say this clearly because mothers are expected to issue a gratitude disclaimer before we admit that anything about motherhood is hard: I chose this life. I wanted this life. I love the people I get to care for.

I am grateful.

I am also tired.

Those things can be true at the same time.

Before I had children, I could spend an entire day on the couch without turning it into a moral crisis. Now, when a free hour appears, my brain immediately begins auctioning it off.

I could switch the laundry.

I could vacuum the floor.

I could answer those emails.

I could order groceries.

I could get ahead.

The problem is that there is no such thing as ahead when you are running a household. There is only the next meal, the next appointment, the next school email, the next load of laundry, and the next object someone needs you to locate.

The list never fully empties. If rest can begin only after everything is finished, rest never begins.

This is the same distinction I keep coming back to when I write about household AI: the task has not left your brain if it has only changed outfits. A prettier list, a better planner, or a prompt you have to rebuild every Sunday still leaves you responsible for remembering that the work exists.

When I use AI to pull dates and requirements out of a school newsletter, the value is not that the list looks nicer. The value is that I no longer have to reread four paragraphs to find the one sentence my family needs.

A bubble bath is not infrastructure

When a mother admits that she has forgotten how to relax, the standard answer is painfully predictable.

Take a bubble bath.

Start journaling.

Buy colored gel pens.

Color-code the family calendar.

Those things are not bad. They also do not solve the problem.

That advice treats overload like a mood. It assumes the mother needs help feeling calmer while leaving every source of the pressure exactly where it was.

Now she has the laundry, the groceries, the school forms, the appointments, the bills, the emails, and a new nightly journaling routine she is somehow supposed to remember.

A bath cannot hold the daycare sick-day policy.

A journal cannot notice that soccer overlaps with a dentist appointment.

A set of gel pens cannot compare this week’s grocery list with the pantry and the household budget.

Self-care cannot compensate for a life that requires one person to function as its entire memory system.

The answer is not to become better at relaxing under the weight. The answer is to reduce the weight.

I did not need permission to rest. I needed fewer open loops.

I have built seven household agents and a household operating system because I got tired of using my brain as the storage system for five people’s lives. But the point is not the number of agents. The point is the number of open loops they close.

Monica, my calendar agent, is the clearest example. She reviews my email twice a day, keeps the calendar updated, and flags what is urgent. At 9 p.m., after the kids are asleep, she prepares the next day’s household briefing: the weather, what the kids need to wear, where everyone needs to be, what needs to be prepared, and what larger events need to stay on my radar.

That is what I mean by infrastructure.

Not replacing people. Not automating love. Not turning motherhood into a technology project.

Replacing remembering.

That describes the system I built for my own household. It is not a built-in feature I can promise every AI account has. The starter handoff later in this post does not connect to Google Calendar, read your email, or run automatically. You will give it the schedule yourself.

Before connecting household information to any AI tool, confirm what the tool can access, what permissions you are granting, how it stores your data, and whether the automation you want is actually supported.

The work did not disappear. My attention came back.

The proof was Spider-Man

Last weekend, I sat down with my kids and watched Spider-Man.

I watched the whole movie.

I did not scroll on my phone looking for something productive to do. I did not mentally inventory the rest of the house. I did not spend two hours physically beside my children while my brain ran laps through everything waiting for me.

I was there.

That sounds small until you have spent years being physically present and mentally somewhere else.

Presence is not measured only in hours. It is measured in how much of you is actually available during those hours.

The systems did not relax for me. They gave me enough mental space to practice relaxing again.

That is the muscle.

Not the ability to ignore your life. The ability to trust that your life will not collapse because you stopped monitoring it for two hours.

What is the first household open loop you would hand off?

Start with the calendar

Do not begin by trying to build seven household agents.

Start with the calendar.

The calendar sits at the intersection of almost everything a family has to remember: appointments, school events, sports, work, deadlines, transportation, clothing, gear, and what needs to be prepared the night before.

Do not begin by connecting anything. Copy tomorrow’s schedule into the AI tool you already use and ask for one screen containing: who needs to be where and when, what needs to be worn, packed, signed, or prepared, any deadlines or conflicts, what you should do tonight, and what information is missing.

Tell it not to guess. If you want weather included, provide a verified forecast yourself.

That is not the finished system. It is the first handoff. Once you know what your household briefing needs to contain, you can decide whether a supported, permission-reviewed connection or automation would remove enough repeated work to be worth building.

The goal is not to build impressive AI.

The goal is to remove one open loop from your brain before tomorrow begins.

Want the build, not just the argument?

Paid Mainframe Mom members get the deeper systems, agents, and starter kits I use to move recurring household work out of my head. The current build is the Marge Starter Kit, which helps you create a meal-planning agent that remembers your household between conversations.

Become a paid member

You should not have to earn an afternoon

I do not want another assignment called self-care.

I do not want to become better at squeezing relaxation between loads of laundry. I do not want to buy a prettier planner so I can document all the reasons I am too busy to use it.

I want a life that stops requiring every task, deadline, question, and decision to pass through my head.

I want to sit down for Spider-Man while the people I love are still small enough to ask me to watch it with them.

That is not laziness. It is not wasted time. It is one of the reasons to build the systems in the first place.

You do not need to become better at carrying everything. You need fewer things to carry.

What is the first household open loop you would hand off?

Leave a comment

Paid members get the systems, templates, and behind-the-scenes builds, not just the ideas.

Subscribe now

Download the Marge Starter Kit and build a meal-planning agent that remembers your household between conversations.

Marge is the name and character I gave mine. You can name yours anything, choose her personality, and teach her how your family actually eats.

The core setup requires no terminal, no coding, and no command-line work. You will install an app, move a folder, select it, and type one setup shortcut.

On the normal path, in about 15 minutes, your agent will know who eats at your table, who is picky, your reliable go-to meals, your grocery budget, where you shop, and which day you plan the week. If Claude needs to install Git for Windows or you hit a troubleshooting step, give yourself a little longer.

Before you finish, you will test her memory before you trust her with a real meal plan.

Yesterday, I wrote about the five words that can turn I’m not technical into a skill you can learn.

Today, I want to talk about what happens after someone convinces a woman that she is a “non-techie.”

They sell her an outcome that requires a system. Then they hand her a prompt pack.

She was promised, “Stop thinking about what to make for dinner every night. AI will plan it for you.”

What she received was a document she must save, remember to open, copy into AI every Sunday, answer, find the follow-up prompt, paste that, review the output, and keep pushing through the sequence until a grocery list eventually falls out.

It may help. It does not remove the remembering.

She was sold a bike and handed training wheels.

A prompt pack tells a busy woman how to repeat a task. A real system remembers the household and carries context forward.

A real system can still have boundaries. The honest version tells you what it handles, what waits for you, and what requires an optional add-on.

A prompt is useful. It is not the system.

I am not arguing that prompts are useless. A good prompt can teach you how to ask better questions, create a faster first draft, or show you what AI can do.

The problem starts when a manual sequence is marketed as a system that runs itself.

If you still have to locate the instructions, re-enter your household information, and manage every follow-up, the task has not left your brain. It has changed outfits.

The setup shortcut you will type today is allowed inside a system. The difference is that you type it once to create your agent’s identity and build her household memory. You do not retrieve it every Sunday to remind her who she is or how your family works.

My Marge is the full build. This kit gives you the foundation.

My own meal-planning agent is Marge, the chef inside my larger Household Mainframe. I named her after Marge Simpson because I wanted someone warm, funny, practical, and capable of keeping a chaotic family fed. That is my character. Yours does not have to be Marge.

Your agent can be calm and efficient, chatty and funny, blunt about the grocery budget, or gentle with a family that is tired of meal-planning guilt. You choose her name, her voice, and what kind of help feels useful in your house.

My Marge knows our soccer schedule, that we need a fast dinner on the nights we get home around 7 p.m., that my husband likes to grill on weekends, and every food my middle child refuses to eat.

My full household version goes further than this starter kit. It runs in the cloud, uses connected household information, and sends the weekly draft to my phone. I review the plan, give her edits, and retain every purchasing decision because money is a permanent human gate in my household.

The starter kit is not the complete cloud system I run in my house. It gives you the part underneath it: a permanent household folder, an identity and personality you control, memory that carries between conversations, and a meal-planning process shaped around your family.

On day one, you open your agent when you want her. She already knows the family, responds in the character you created, and absorbs any notes waiting in her folder. She does not initiate work while Claude is closed. If you later want a scheduled nightly check-in, the kit includes a separate optional add-on. Nothing essential is lost if you skip it.

The Marge Starter Kit does not pretend the Claude desktop app can perform background work it cannot perform by itself.

Yesterday, I made the larger case against selling women the identity “I’m not technical.”

Here is the part that matters for this build: women already tend to rate their technical ability below what their performance supports. Researchers who observed adults using the web found little difference in actual ability but significantly lower self-assessments among women. A study of computing students found the same pattern among women and men earning equal grades.

Women do not need marketing that turns that confidence gap into a customer identity. You can install a persistent AI system, teach it about your household, test its memory, and understand its automation boundary without knowing how to code.

Paid readers: today you get the foundation that makes Marge remember

The Marge Starter Kit includes:

• The persistent foundation behind my Marge: her folder, instructions, customizable personality system, and built-in meal-planning process

• One guided /setup interview where you name your agent, shape her personality, and teach her how your family actually eats

• Persistent memory that stays inside the kit folder

• Windows and macOS setup instructions

• Windows troubleshooting for Git installation and the claude is not recognized error

• An optional nightly check-in add-on for readers who want background scheduling

The kit gives you the infrastructure. You give it a character, teach it your household, and retain the judgment.

Become a Paid Subscriber and Build Marge

Before you start

You need:

• A Windows or Mac computer

• A paid Claude Pro account. The free plan does not include what this kit needs.

• About 15 uninterrupted minutes on the normal path. Allow longer if Claude needs to install Git for Windows or you need the troubleshooting guide.

For the core kit, you do not need a terminal, GitHub account, API key, or coding experience. You will type one setup shortcut exactly as shown.

Read more

What This Place Is

Mainframe Mom documents one thing: handing the invisible load of running a household to AI, so the human in the house can stop being the operating system. I install a system in my real house, then I write down exactly how, so you can install it in yours. No coding. No new app to babysit.

I’m a civil engineer and a working mom. That’s the whole credential.

If You Only Read 3 Things

1. I Was on the Floor With My Kid and Still Not There — the post that explains why any of this exists.

2. The Deadline Is in Paragraph Four — a two minute move you can steal today.

3. Women Were the First Programmers. Then They Told Us to Leave. — the history that reframes who tech belongs to.

The Mental Load

For the week you feel like the only one holding the map.

• The 8:40 PM Walk-Through — the nightly shift nobody assigned, but you keep clocking into.

• Mental Load Is Not a Personality Trait 🔒 — remembering everything is labor, and labor can be handed off.

• My Marriage Got Stronger After I Stopped Asking for Help 🔒 — what changed when the asking stopped.

AI That Actually Helps

The installs. Real prompts, real routines, from my kitchen to yours.

• I Never Wrote Code, but I Use Claude Code for 90% of My AI Work — the five-second test for knowing when AI should become an actual workflow.

• The Deadline Is in Paragraph Four — school newsletters hide their deadlines in prose. One move turns them back into a dated list.

• Using AI as a Chatbot Is the Tutorial Level 🔒 — stop asking AI questions, start assigning it jobs.

• My AI House Wife, Part 1: Your To-Do List Was Never the Solution 🔒 — where the whole system started.

• I Built an AI to Run My House, Zero Coding 🔒 — meet Etta, the system that runs my week.

Being All the Way Here

The point of the whole system.

• I Was on the Floor With My Kid and Still Not There — present from the outside, running a call center from the inside.

• I Gave My Kid a Face That Looked Like Listening — the tab that opens mid-conversation.

You Were Always Ambitious

• 5 Words That Turn “I’m Not Technical” Into a Skill I Can Learn — why a knowledge gap should never become your identity.

• Women Were the First Programmers. Then They Told Us to Leave. — you belong in this space. Historically. Literally.

• Warm Is the Flex in a Room Full of Grey. — the jargon is a keep-out sign, and warm wins anyway.

Building in Public

Every Sunday, I publish what shipped, what sold, what flopped, and what I’m changing next in the Build Log. The first Sunday of every month is the Scoreboard: subscribers, revenue, lessons, and the next target.

The free stuff: The Free Toolkit holds a dozen free resources, each doing one job on one page.

The member stuff: Paid subscribers get the exact prompts, templates, blueprints, and installs, plus $10 off The Household Mainframe.

The founding tier: Founding Members get The Household Mainframe included and directly fund the next systems I build, test, and hand back.

If any of this sounded like your kitchen, subscribe. One tab, closed.

Subscribe now

Making technology accessible should expand what women believe they can do, not sell them a smaller identity.

Mainframe Mom is for ambitious working moms who want AI and better systems Free every Sunday. Deeper builds every Thursday. The five-word reframe is at the end.

Subscribe now

“Give me the guy engineer.”

I was standing on a job site as the civil engineer representing a local government agency. A representative from the local Department of Health was there, too. He refused to speak to me.

Not politely. Not subtly. Not in one of those situations where you spend the drive home wondering if you imagined the whole thing.

“I’m not saying anything to you,” he said. “Give me the guy engineer.”

He had already decided what technical competence looked like, and it did not look like me.

Women have spent decades trying to climb out of that box. We are still climbing out of it.

Recently, I stopped on an Instagram Reel aimed directly at “an ambitious non-techie who’s tired of AI feeling like an exclusive club.” The creator’s promise was to build a community where AI feels practical, fun, and useful.

There is nothing wrong with building that community. There is nothing wrong with teaching beginners. What caught me was the identity being handed to the viewer before the teaching even began: non-techie.

That Reel is one specific example, but I see variations of the same framing repeatedly in my own feed:

“Learn AI as a non-technical person.”

“Here’s how I edit my Reels as a non-tech person.”

The most generous reading is that a creator using this line wants to sound relatable. She wants the woman watching to think, Oh, good. This won’t be over my head. She is trying to communicate that her course, template, or workflow is simple, practical, and mercifully free of tech-bro language.

That intention is not the problem.

The sales mechanism is.

Before the offer can sell the solution, the framing gives the buyer a limiting identity: You are not technical. I am not technical. Luckily, I found a system easy enough for people like us.

The creator may not believe her audience is incapable. But the framing still asks the buyer to accept a smaller story about herself before the offer can rescue her from it.

“You’re not a tech person” is not a fact. It is a sales pitch that turns a skill gap into an identity.

A knowledge gap is not an identity

There is a difference between saying, “No coding experience is required,” and saying, “You’re not a tech person.”

The first describes a prerequisite. The second describes a person.

This is the same trap I wrote about in “Mental Load Is Not a Personality Trait.” Women keep getting handed character labels for things that are actually systems problems or skills they were never taught. “Non-tech person” is the business-building version.

There is a difference between promising, “I’ll explain every step clearly,” and marketing something as easy enough for a “non-techie.”

The first makes the teaching accessible. The second makes the student smaller.

“Technical” is not a gene some people are born with. It is not a personality type. It is not a private club whose members arrived knowing how to connect a domain, build a website, edit a video, format a digital product, or troubleshoot a broken automation.

Those are learned skills.

Sometimes you have learned them. Sometimes you have not learned them yet.

That final word matters, especially for the woman staring at a blank website template at 11 p.m., wondering whether she is capable of building the thing she can already picture in her head.

She does not need another person confirming that she is “not technical enough” to build it. She needs someone to teach her what she needs to know.

Simple can still be technical

The strongest defense of “non-techie” marketing is also the most reasonable one: creators want beginners to know their teaching will be easy to follow.

Good. It should be.

I have no interest in defending confusing instructions, unexplained jargon, or courses that make basic workflows sound more complicated so the person teaching them can appear smarter. Clear teaching is good teaching.

The opposite of tech-bro teaching is not “non-technical” teaching. It is humane technical teaching: clear language, enough context, and no performance of superiority from the person who already knows where the buttons are.

But things can be simple and easy to follow and still be technical.

You do not make technology accessible by removing the technical parts. You make it accessible by explaining those parts well.

Teach her what the platform is doing, not only which button to press. Teach her why the workflow is built that way. Teach her what can break, what the error message means, and how to investigate it. Teach her enough that when the interface changes next month, she does not assume she is the problem.

Teach her everything. The technical aspects may be the most important parts.

If a course gets a woman to the promised result only as long as every screenshot still matches her screen, it has not made her more capable. It has made her dependent on the next tutorial.

That is not accessibility. That is a funnel with excellent lighting.

You are allowed to figure it out

My life revolves around technical work. I am a civil engineer, and I genuinely enjoy technology. But that does not mean I automatically know how to do every technical task my business requires.

I am not a web developer. I am not a graphic designer. I am not an accountant.

I am resourceful.

That is the skill that matters.

If I do not understand part of an accounting workflow, for example, I can use AI to translate the unfamiliar terms, map what information belongs where, and help me identify the next question to ask. That does not make me an accountant, and it does not mean I should never hire one. AI is simply one resource I can use to move from I don’t know this to I understand my next step.

You do not need to become an expert in every platform before you start a business. You do need to believe you are capable of learning enough to build, evaluate, troubleshoot, and know when professional help is warranted.

There is an enormous difference between those two beliefs:

I don’t know how to do this yet.

I’m not a tech person.

One identifies the next skill to learn. The other closes the door before you touch the handle.

Stop selling women the box

What bothers me most is that I notice this language so often coming from women selling to other women.

We know what it feels like to have our competence questioned. We know what it feels like to be talked over, underestimated, or treated as though the technical part of the room belongs to somebody else. Some of us have stood on job sites while men demanded “the guy engineer.”

Why would we repeat that message inside our own businesses?

We can be warm without being patronizing. We can teach beginners without calling them incapable. We can make complicated things clear without pretending the people learning them are fundamentally “non-technical.”

The woman building a business may be inexperienced. She may be overwhelmed. She may be trying to connect a domain while convinced everyone else received a manual she somehow missed.

She is not dumb.

She is not disqualified.

She is learning.

The next time your brain offers, “I’m not technical enough to build this myself,” replace it.

Try: “I can figure out the technical aspects I need to build this.”

Then find the first part you do not understand. Ask the embarrassing question. Look up the term. Try the button. Try again.

Technical confidence is not knowing everything before you begin. It is trusting yourself to stay in the room when you do not know yet.

Stop trying to put yourself in the box we have spent decades trying to get out of.

You are smart. You are capable. And you can be technical.

The five-word reframe: I don’t know how. Yet.

Not knowing is temporary.
“Non-tech person” turns it into an identity.
You are capable of learning the next piece.

What is one technical task you are ready to stop using as evidence against yourself?

Leave a comment

Practical AI and better systems for carrying less, every Sunday. Free.

Subscribe now

For ambitious moms who want AI and better systems to carry more of the load. One sharp, practical idea for reclaiming your mental bandwidth, delivered every Monday. At the bottom: the five-second test I use to decide between Chat and Code.

Subscribe now

If It Might Become a Workflow, Stop Starting in Chat

Claude Code sounds like a place for coders. I have never written a line of code.

On June 27, I opened Claude Code and asked it to review a proposal I had written for work and identify areas of risk.

It reviewed the actual document. Then I could keep going, work through the risks, have it update the Word file with my edits, and save the finished document to my Google Drive.

The conversation did not have to end with advice. It could continue into the actual work.

This would have been a fairly ordinary evening if I had not spent months being terrified of Claude Code.

Its interface is a little scary. The concept of “coding” is scary. The word Code is right there in the name, aggressively suggesting that you should know things.

And that is coming from someone with an engineering degree.

I assumed Claude Code was for people who built software. People who understood terminals and commands and whatever else was happening inside that very serious-looking black box.

I have never once written a line of code.

I now use Claude Code at least 90% of the time.

Where does the result need to live?

Chat is fantastic when I want an answer.

Explain this letter from school. Help me think through this decision. Give me five dinner ideas using the food in this picture.

The answer can stay inside the conversation because the conversation is the final destination.

But I kept using Chat for work that needed to go somewhere else.

I would ask AI to review something, then manually transfer its feedback into the document. I would ask it to create information, then copy that information into another program. I was using an incredibly capable tool and still volunteering to become the human conveyor belt between every step.

That is the same trap I wrote about in Using AI as a Chatbot Is the Tutorial Level. If AI gives you an answer but you still have to carry that answer through every remaining step, the work never actually left your plate.

Claude Code gives the answer somewhere to go.

Chat gives me an answer.

Code helps me turn the answer into work.

There is friction. There is not coding.

I am not going to tell you that the first five minutes feel exactly like opening a normal chat window.

The interface is unfamiliar. There is some setup. You may need to approve permissions before it can access or change anything on your computer.

That friction is real.

But once you are inside, you communicate with it the same way you would communicate with your smartest, most capable best friend.

You explain what you want in plain English.

You do not need to know the command Claude Code will use. You need to know the outcome you want.

If you are nervous, give it the smallest possible job:

Create a file named todo.txt and list three ways an AI tool can help a beginner automate daily computer tasks.

Watch it create an actual file on your computer.

Then ask one follow-up:

Add one simple example beneath each idea.

The first instruction produced a real artifact. The second changed it. You did not have to copy a response out of a chat window or manually update the file yourself.

You directed the work.

Only after that tiny win do you need to think about what else becomes possible.

Eventually, that might look like my household system, where a meal-planning agent can use budget information produced by my bookkeeping agent. Two pieces of the household can work from the same information without me carrying numbers from one conversation into another.

But your first workflow does not need multiple agents.

It needs one text file.

The five-second test

Before you ask AI something, take five seconds to decide what you want to happen after it answers.

If the answer can stay inside the conversation, start in Chat.

If the answer needs to become a document, spreadsheet, folder of files, or process you will repeat, start in Claude Code.

Do not automatically start with the interface that feels most familiar and then volunteer to carry the result the rest of the way.

Start where the final result will be most useful.

I spent months avoiding Claude Code because I thought I needed to become more technical before I could use it.

I didn’t.

I just needed to stop standing outside the door and assuming the word Code meant I wasn’t invited.

If the answer can stay in the conversation, start in chat.

If it needs to become a file or workflow, start where the work will live.

Tell me in one line: What is one task you keep starting in chat that should probably become a workflow?

Leave a comment

Get the Next Mainframe Minute

One sharp, practical AI idea for reclaiming your mental bandwidth, delivered every Monday, free.

Subscribe now

Collisions are cheap in August. In September, they actually cost you.

The ambush is never the event. The ambush is the intersection: the school’s schedule crossing your family’s actual life at an angle nobody checked.

Soccer practice is not an emergency. Neither is a pediatrician appointment, the first day of school, or a work meeting.

Put them in the same two-hour window with a toddler who has reached the end of her thirty minutes of patience, though, and suddenly somebody is eating a granola bar in the car while you explain to your boss why you cannot make the meeting you accepted three weeks ago.

That is September.

September is a hard cutover

Summer is chaotic, but it is flexible chaos. Bedtime drifts. Dinner can be late. Camp schedules are annoying, but they do not usually arrive with picture day, bus times, spirit week, homework, practice, school forms, class assignments, and a pediatrician’s calendar stacked on top of them.

September is different. September is a hard cutover from one operating system to another, except nobody gives the household an implementation plan.

This year, soccer starts on the first day of school. I have two boys in two different age groups with two different schedules. Two weekday evenings are now practices. Every Saturday morning is a doubleheader. For roughly eight weeks, I cannot casually accept an evening meeting and assume I will figure it out later.

Those two practice nights also need a very specific kind of dinner: super quick, late, and composed entirely of foods each child will actually eat. The cleats need to fit. The shin guards need to be found. The balls need to be in the car. Water bottles need to exist in the same physical dimension as their lids.

And then there is my one-year-old, who can happily watch practice for about thirty minutes before she is personally done with organized sports. So the schedule is not just “soccer at 5:30.” It is transportation, dinner, gear, work boundaries, toddler containment, and the emotional regulation of everyone involved.

The event is one line on a calendar. The intersection is the whole job.

The collision-finding job

Back-to-school already has its own structural reset: haircuts, shopping, supply lists, class assignments, updated routines, and the annual emotional roulette of whether your kid’s best friend lands in the same class. We have lived through the best-friend-not-in-class devastation before. I know it may happen again. Knowing that does not make the dread smaller. It simply adds another invisible thing I am preparing to hold.

Then I noticed my eldest’s annual checkup is on the second day of school, in the morning, directly at bus time.

That appointment cannot be casually moved. Our pediatrician books roughly three months out. So the collision is not “checkup on Thursday.” The collision is: find childcare for the other kids, make them miss part of the second day of school, or move an appointment deep into the future and absorb whatever that changes next.

Found in August, that problem costs a few texts and a decision. Found the night before, it costs panic, a favor, missed school, missed work, or all four.

Why “just put it on the calendar” is not enough

I once skimmed a school email and missed the memo that it was “tacky tourist” day. My kid arrived dressed like a completely normal child while everyone else looked ready to board a cruise ship in 1997.

He was the only one who missed it.

This still visits me at 2 AM, because motherhood is beautiful and the brain is a deeply unserious filing cabinet.

The event technically existed. The information had been sent. The failure was not that I lacked a calendar. The failure was that the meaningful detail was buried in one document while the rest of our life lived somewhere else. I had to notice the sentence, understand what it meant, connect it to the right child and date, determine whether anything had to be bought, and surface it again at exactly the right time.

A calendar can store “Friday: spirit day.” It does not automatically ask whether you own a Hawaiian shirt, whether it still fits, whether the reminder needs to happen before Target closes, or whether another child has something equally weird happening that morning.

That is the part we keep calling organization. It is actually intersection detection.

Two datasets. One exhausted human processor.

The school publishes its half of the data in a tidy PDF every August. Your family’s half sits across your phone, work calendar, pediatrician portal, sports app, custody schedule, daycare messages, and a text from another parent you were absolutely going to answer.

And the only place those datasets ever meet is behind your eyes.

That is why you can have every date entered and still feel the low-grade “I’m going to miss something” dread. Your brain knows the calendar is not the finished product. It knows it is still responsible for comparing every new event against the rest of your family’s life.

The problem is not that you are bad at calendars. The problem is that the calendar is holding events while you are still holding the consequences.

And this particular job is shaped like a machine’s job: two documents in, intersection list out, no judgment calls.

Behind the paywall is the whole paid deliverable: a printable one-page Season Map, the three-bucket Collision Check that fills it, and the copy-and-paste prompt that turns two messy calendars into one short decision list.

Read more

Subscribe now

P.S. Paid subscribers get the Offload Audit worksheet which finds the five parts of your mental load interrupting you most.

There's a ten-minute prompt at the bottom of this one. Steal it.

For the last month I’ve been posting articles on Substack that were written by AI.

All of them. I read them. I edited them where they needed it. But AI did the work, and there are a few I couldn’t tell you what I said.

I have 26 subscribers.

The day you’re reading this is my 35th birthday. August 16th. Leo energy. I’m not afraid of a whole lot, which is the only reason I’m telling you any of this.

I also want to monetize this. People get weird about that part, so I’ll say it plainly on my birthday. I built a Substack because I have ambition and I want it to make money someday.

This may work. It may not. This is me building it in real time.

Why would you hand your writing to a robot?

I work full-time as a civil engineer. I love my job. I went to school for it. Eight years, two degrees, and six figures in debt. But I love it, and I love using that part of my brain. I will never not want to use that part of my brain.

I also write in 15-minute increments. Two evenings a week, at most. Work from home, full-time, three kids. You can imagine how crazy my days get.

So I handed the whole thing to AI. I figured I’d tell it what my brand was about and that would be good enough.

But that’s not why I created a Substack in the first place.

What 26 people actually bought

There are people who will read that number and say: you got 26 subscribers and you weren’t even writing your posts.

I think that’s fucking awesome.

Every single one of those is a person. That’s 26 people who wanted to hear what I had to say badly enough that they were willing to let AI say it for me.

One of them was the article below. Go read it and tell me if you can feel where I’m missing.

I’m so freaking thankful for every one of those 26 people.

But AI was never the problem.

The detail nobody would have thought to brief

It’s 10:02 on a Friday night. My kids are in bed. I read them a bedtime story, and it was drafted by AI, because I’m crap at making up children’s stories.

My kids loved it. They loved Gus the Dinosaur helping them make a rainbow. And they both laughed hysterically at the part about their sister riding in a wagon with a cracker in her hand.

AI knows I have a daughter. It doesn’t know she’s obsessed with crackers right now. It doesn’t know the cracker was the funny part of tonight.

That’s what a reader gets from a human.

What it cost my kid when I forgot

Last winter I forgot to register my oldest for wrestling.

It cost him a season. He still talks about it. I had to sit there and explain to him why I failed.

So I built the thing that makes it impossible to happen again. If a registration form comes through my email, an agent scrapes it, puts it on my calendar, puts it on my to-do list, and reminds me relentlessly until it’s done. If it comes home in a backpack, I take a picture and send it over. Same thing.

That’s AI doing exactly what I want AI to do.

I did the same thing for this newsletter. Everything around my draft is now one system. I'm calling it Creator Mainframe. The brainstorming that gets you off a blank page, the title, the SEO, the flat-spot check, the notes that go out after, and a log of what each post actually did. This article is the first thing that came out of it with the body still mine.

And I didn’t think to tell AI any of it until right now, sitting here with my brain in motion. The wrestling season lives in no file I have ever handed a model. Not in a voice DNA doc. Not in a storytelling prompt.

AI is great at writing. It’s still not me.

The $565 I didn’t know I was paying

I just started using AI in my household. I have a household agent. Her name is Doris. She’s my bookkeeper.

She found $565 a month in subscriptions I had no idea existed.

That’s embarrassing as fuck (yet, here I am writing about it #LeoEnergy). And AI wouldn’t have written about it, because AI didn’t know it mattered for context in one of my articles until I did.

So where's your line?

The draft stays with me. Everything around the draft goes.

Here’s what that looked like tonight, with this article.

I gave it the title. I gave it the SEO, which is the shit I don’t care about. I gave it the photos. Then I handed the whole thing back and asked it to tell me where the piece falls flat and which parts land, so I’d know what to cut.

Every word of the body is mine.

Where the line sits is different for everybody. I’m fine giving AI my visuals. If you run a photography business, you probably shouldn’t, because that’s your whole leverage.

I saw a woman on Instagram this week upset about her carousel getting flagged for AI. The thing that flagged it was the Canva background remover. She called Meta to confirm it. She’s still whining about it.

Girl. You’re using AI. Keep using it.

Stop purposefully making your life more difficult just to follow suit that AI is the villain.

Let it serve you. Keep your voice out of its hands. Hand it the things you find mundane, the things you don’t like doing, the things you’re honestly just not good at.

Your turn tonight

I’m challenging you to use AI in your creative process. Give it the job of finding your article’s weak spots.

Open it up. Paste in a draft you’ve got sitting around. Then paste this:

Ten minutes. Free.

Come back in a month and I’ll tell you what the number is.

The draft stays with me. Everything around the draft goes.

Where that line sits is different for everybody.

Mine is the body of this post.

What’s the one thing you’d never hand to AI in one line?

Leave a comment

Everything around the draft is now one system for me. The brainstorming, the title, the SEO, the flat-spot check, the notes, the log of what each post did. I'm building it as Creator Mainframe, and this post is the first thing that came out of it with the body still mine. Waitlist is open.

P.S. Thursday I'm writing about the anti-writer's-block piece of it. How a half-baked idea becomes an article I actually want to read and want to write.

Every book, every course, every 5am-club influencer starts from the same quiet assumption: there’s a better version of you in there somewhere, and if you just optimize hard enough, you’ll finally become her. Wake up earlier. Journal more. Meal prep on Sunday. Build the perfect morning routine. Become the woman who has it all handled.

I bought in for years. I have the planners to prove it. There’s one sitting in my kitchen right now, opened to a fresh week, highlighter in hand, because surely THIS was the system that would fix me.

Here’s what nobody in that aisle will say out loud: they keep selling you a better version of you because if you ever stopped fixing yourself, you’d notice the real problem. You are the only system running the whole house.

The self-improvement machine only makes money if the fix is always internal. The second you point the finger at the setup instead of yourself, the sale is over. So it never does. It tells you you’re scattered, undisciplined, not trying hard enough, one habit-stack away from finally keeping up. It never tells you that keeping up was never a personal failing. It’s a math problem.

Because look at what every one of those systems actually assumes. The 5am routine assumes you have a quiet hour nobody else needs. The cold plunge and the gratitude practice assume you have spare capacity lying around, waiting to be optimized. The color-coded productivity method assumes the bottleneck is your focus. None of them assume what’s actually true: you’re doing the job of three people, you’re the only one who knows when anything runs out, when the forms are due, what everyone eats, and there is no spare capacity to optimize. There’s just you, doing all of it, being told to do it better.

I’m an engineer. When a system is overloaded, you don’t tell the one processor carrying every task to have a better attitude about it. You look at the load. You move the work somewhere else.

So I stopped. I stopped trying to become a faster, calmer, more disciplined version of the person doing three unpaid jobs. That woman didn’t need a better morning routine. She needed fewer jobs.

I handed the jobs themselves to Etta. The AI running my house doesn’t need me to wake up at 5am to be productive. It orders the groceries, tracks what’s running low, books the appointments, remembers the field trip form that lives in nobody else’s brain. Not so I can cram more into the space that frees up. That’s the trap running the other direction, and I’m not doing that one either.

I offload everything else precisely so I can stop optimizing me.

That’s the whole point. I don’t hand tasks to Etta so I can become a more impressive person. I do it so I can sit on the floor and actually hear the end of a story my kid is telling me, instead of nodding while I silently draft a grocery list behind my eyes. Presence isn’t a productivity metric. It’s the thing all the productivity was supposed to buy and never did.

The self-improvement industry wants you to believe you’re a work-in-progress, a project that’s never quite finished, always one purchase away from fixed. You’re not. You are a person doing the work of three, and the thing that needs fixing has never been you. It’s the load.

Stop optimizing the processor. Move the work.

I’m not a better version of myself. I just stopped being the whole operating system.

How many planners did you buy hoping to fix a problem that was never yours?

Leave a comment

P.S. If a friend sent you this: the letters are free. Subscribe and the next one arrives on its own. Which, you'll notice, is the whole philosophy around here.

Subscribe now

P.P.S. You can’t hand off a job that isn’t written down. The Household Mainframe is the writing-it-down part — a twenty-minute interview that moves the operating knowledge of your house out of your skull and into a document any adult can run. $37, done once.

Grab the Household Mainframe

Monday, 6:14am, before anyone else in the house is awake. My phone buzzes with a brief I didn’t write that morning. Etta did, while I slept. The week’s appointments, the two things that will fall through the cracks if nobody watches them, the one kid who has a form due Thursday. I didn’t ask for it. It just showed up, because I built it to.

That’s the part nobody explains. The gap between people who think AI is overhyped and people whose lives quietly changed was never a skill gap, and there are no secret magic words. One group is still typing questions into a box. The other group stopped asking and started delegating.

Prompting is you doing the work with a faster typewriter. You show up, you ask, you read the answer, you act on it, and tomorrow you do all of that again. A great prompt gets you a great answer, but you had to be there for every step of it. It feels productive because the outputs are good. It’s still you carrying the whole thing, in a conversation you have to keep restarting, forever, at the exact moment your brain is already full.

Building is different. Building is handing over the whole job and walking away. I set up Etta’s standing instructions for the Monday brief exactly once. I told her where to look, what I care about, what to flag, how to talk to me. Now the brief happens whether I remember it exists or not. I’m not in the loop anymore, and that is the entire point. The difference between cute and useful is whether the thing needs you sitting in the chair for it to run.

Most people never cross that line, and intelligence has nothing to do with it. The people teaching them stop at the line on purpose.

Think about what’s easy to sell. A prompt library. A swipe file of magic phrases. A carousel of ten prompts that will change your life. It’s cheap to make, it’s easy to consume, and it makes you feel like you got somewhere without changing anything about how you actually work. The build, the real one, is harder to package into a screenshot. So the gurus hand you the training wheels, call it the bike, and let you wobble around the driveway feeling like you’re riding.

You’re not underwhelmed with AI. You’re underwhelmed with the training wheels, and no one told you there was a bike.

I want to be honest about what the build actually is, because I’m not going to sell you a fantasy where you say one clever sentence and your life runs itself. Building a system means deciding what job you’re handing off. It means writing down how you’d do it if you had to explain it to a new hire, because that’s basically what you’re doing. It means the first version breaks, and you fix it, and the second version breaks somewhere dumber, and you fix that too. It’s a couple of unglamorous evenings, not a montage.

But here’s what you get on the other side of those evenings. A task you hate, done, on a schedule, without you. Not a better answer to a question you had to ask. An employee you don’t have to pay, doing the thing while you’re asleep or on the floor building blocks with your kid instead of silently drafting a to-do list in your head.

So if your AI only works when you’re sitting there talking to it, you don’t have a system. You have a smarter search bar, and a smarter search bar is still a tab you have to keep open. The whole game changes the day you build something that runs when you’re not looking.

Stop typing questions into the box. Go hire the intern who never sleeps, then teach it your job once and never do that job again.

Behind the paywall is the actual build: the skill file copy-paste ready, Etta’s standing instructions verbatim, and the two things that broke before Monday 6:14 ever worked.

Read more

Not on purpose. Nobody hands me a schedule. But somewhere between the last cup of water and the couch, the walk-through starts, the same way it starts every night. Tomorrow, from the top. Who wakes up first. Who has to be out the door by when, wearing what, carrying what. Which adult is on drop-off, and whether he knows that. The form. The snack that can’t have nuts. The thing I told the teacher I’d send in.

I run tomorrow like a dress rehearsal for a play where I’m also the stage manager, the prop department, and the only one who has seen the script.

Last night’s run, so you know I’m not being poetic:

Launch at 6:35 tomorrow, not 6:45, because the toddler goes back to daycare first, then the boys to camp. Water bottles in both camp bags. Plus an extra bottle of propel since it’s supposed to be hot. I have meetings at 9:30 AM, 11 AM, 1 PM, and 2 PM. I need to start the laundry before my 9:30 so I can put it away after my meeting ends at 12 but before my next meeting at 1 starts. But I also need to clean the bathrooms. Crock pot beef tips for dinner. So between 12 and 1 I need to put away laundry, start dinner, and clean the bathrooms. That will have to be my lunch break for the day. Leave at 3:52 for the 4:00 pickup. Then pick up the toddler afterwards so I don’t have to buckle and unbuckle her from her car seat 32x…

That’s one night.

Here’s what I want to say about that rehearsal, and it isn’t “just let it go.”

The rehearsal works. That’s the trap. The night I walk through tomorrow is the night I catch that both boys need to be at different fields at the same hour, or that the early release lands on the one day neither of us can leave work. The walk-through finds real collisions before they happen. It has saved our family’s Tuesday more times than I can count. Anyone who tells you to simply stop doing it is telling you to unplug the smoke detector because the beeping is annoying.

So I kept doing it. Every night. For years.

And here’s the accounting nobody does: a briefing that runs every single night is a shift. At any job, the person who prepares the daily rundown, checks the schedule for conflicts, and briefs the team has a title and a salary, and when she goes on vacation, someone covers it. At my house the same job ran nightly, unpaid, unassigned, from memory, in the dark. There was no coverage. If I fell asleep before finishing, the rehearsal woke me at 2am to run the rest.

Nobody staffed it. It staffed itself, onto the person who noticed first. That’s how every invisible job in a household gets assigned: not by conversation, but by whoever can’t not see it.

I want to be careful here, because this is the part the internet gets wrong. The villain is not my brain, for refusing to power down. My brain was doing something necessary that nothing else in the house was doing. And the villain is not my husband, who would have gladly taken a list if a list existed. The villain is that the entire operating knowledge of tomorrow existed in exactly one place, and the place was me, and the only time the retrieval system ran was when I lay down.

The exit wasn’t relaxing harder. The exit was moving the briefing.

It started small, with a two-minute version: saying the rehearsal into a box instead of holding it. (I wrote about that here.) But the real shift happened when the briefing stopped needing me at all: when tomorrow started arriving, already walked through, collisions already flagged, before I woke up. Built once. Runs nightly. Doesn’t wake anyone at 2am.

That build is tomorrow’s letter, and it’s the deepest one I’ve written, because it’s the difference between using AI to cope with the load and building something that carries it.

For tonight, though, just this: if you clock in at 8:40 too, the rehearsal is not a character flaw and you are not “bad at relaxing.” You are running an unstaffed shift from memory. It counts as work. It has always counted as work.

The reason it all lives in your head is that nobody ever wrote it down anywhere else. That’s the actual fix, and it’s the first thing I ever built: a twenty-minute interview that moves the operating knowledge of your house out of your skull and into a document any adult, or any system, can run. It’s called The Household Mainframe, it’s $37, and it’s the reason my 8:40 belongs to me now.

The house is quiet. It can stay that way.

Subscribe now

Leave a comment

Nobody scheduled it. You lie there and run tomorrow. Who gets dropped where, and by which adult. The library books. The form that’s due. The thing you’re supposed to defrost, sign, find, or remember to say to the teacher at pickup. You run the whole day like a dress rehearsal, in order, and if you wake up at 2am, the rehearsal runs again from the top.

Here’s the two-minute version of quitting that shift.

Tonight, when the rehearsal starts, open an AI chat instead. Say it out loud into the microphone, or type it with your thumbs, exactly as messy as it is in your head. No order, no complete sentences. “Early release Wednesday, someone has to get him at 1, field trip money, I think the cleats are too small, dentist rescheduled, defrost the chicken.” All of it, until the loop is empty.

Then send one more line:

Turn this into tomorrow’s list, sorted by time. Flag anything another adult has to do, and anything that can actually wait past tomorrow.

What comes back is tomorrow as a list instead of a loop. Paper could have made you a list too, sure. The difference is that you didn’t organize anything. You talked, the way you already talk to yourself in the dark, and the sorting happened somewhere that isn’t you. The “can wait” flag alone is worth the two minutes, because the 8:40 rehearsal never grants extensions. The box does.

That’s the whole move. Two minutes, tonight, and the rehearsal has an understudy.

Tomorrow I’m writing about that 8:40 shift itself: who staffed it, why it exists, and what happened when I stopped being the only one who could run it. Subscribe and it lands in your inbox in the morning, which is to say, after the shift.

Subscribe now

I was at a birthday party, running point on three kids and a snack rotation, when a woman leaned in like she was about to change my life. “You know,” she said, “you could just ask for help.”

I’ve heard it my whole adult life. Ask your husband. Split the list. Divide and conquer. It sounds like relief. It is not relief.

Here’s the part nobody says out loud: asking a person for help hands you a brand-new task, managing that person, and the knowing never moves an inch. Think about what “asking for help” really means in a house. It means I still have to know the thing needs doing. I have to know when it’s due, what it looks like when it’s done right, and whether it got done at all. Then I have to translate all of that into an assignment, hand it over, and remember to check. “Ask for help” is a polite way of saying stay the manager. The task moved. The load didn’t. It just put on a helpful little disguise and sat right back down in my brain.

And delegating to a person who needs reminding is two jobs. There’s the original thing, and then there’s the whole second job of getting a human to reliably do the original thing. Assign it. Re-explain it. Follow up when it stalls. Absorb the “you didn’t tell me” when the field trip form is due tomorrow and nobody else in the house knew it existed. By the time I’ve managed the handoff, I could have done the task four times. So I do. And then people call me “so organized,” which is a nice word for “the only one who’s tracking any of this.”

So I stopped trying to delegate the tasks and started delegating the knowing.

Etta, the AI system I built to run my house, doesn’t need managing. That’s the whole point of her. The pediatrician, the bottle recipe, the field trip date, the day the camp bill is due: the knowing lives in her now, and it stays there whether I’m thinking about it or not. She surfaces things before they become crises. Nothing pings back to me for a decision I don’t need to make. There’s no re-explaining, no checking, no “did you remember to.”

That is the difference, and it’s the entire difference. A person you delegate to becomes another tab open in your head. A system you build closes the tab. One adds a management job. The other deletes it.

I’m not saying your partner shouldn’t do more. He should. Plenty of them should. But “ask him” as the solution to your mental load quietly assumes the load is a chore list, when the load is actually the knowing, the tracking, the remembering, the being-the-only-one-who-notices. You can split the chores down the middle and still be the sole person holding the map. That is a systems problem, and nagging has never once redesigned a system.

You were handed a version of help that still runs on you. The design flaw was built in long before you got there, and it was never yours.

Help that lives in your head is just you, with extra steps.

Below the wall: how the knowing actually leaves your skull. The intake ritual, the rules for when Etta speaks up, and the morning a field trip got handled without me.

Read more

Emergency contact. Pediatrician’s name and number. Insurance carrier and member ID. The two adults authorized for pickup. The allergy line. The grandparent’s cell. You have produced these answers, from memory, on every form, for every kid, every year, sometimes four times in a single packet.

Nobody else in your house can pass this exam. That’s the part worth sitting with for a second. If the school called and you were on a plane, the household would be calling you at the gate, because the answer key to your family’s own logistics lives in exactly one place, and the place is you.

Same story, worse stakes, for the first week of school. Who gets which kid, at what time, through which door. Which bus number. What the aftercare pickup password is. Whether Wednesday is early release, because of course one day is. Right now that information exists as a briefing you deliver, out loud, nightly, to whichever adult is on duty, and re-deliver when they forget, because spoken words are a terrible storage medium.

At work nobody operates like this. Repeated questions get a reference document. It gets written once, by the person who knows, and then the knowing is a file instead of a person.

So this weekend’s upgrade is exactly that, twice. Two pages. Twenty minutes total, most of it just transcribing what’s already in your head. One page answers every form the school will ever send. One page runs the first week of school without you narrating it. I built both as fill-in templates, and the second one has a rule attached that is the actual product: once it’s written, you are allowed to say “it’s on the sheet.”

Below the wall: both templates in full, the twenty-minute fill-in order, where each page lives so the whole rotation can see it, and the one rule that keeps them from quietly becoming your job again.

Read more

She’s at the pool. She’s at the lake house, the backyard sprinkler, the good long evening where nobody has anywhere to be. Physically, fully there. Sunscreen on every kid.

Mentally she’s in September. She’s running the supply list. She’s wondering when the aftercare signup opens. She’s doing the quiet math on whether the pediatrician can get him in for a sports physical before practice starts. She is, as far as anyone can see, watching her kids play. She is actually at a desk in her own head, managing a month that hasn’t happened yet.

I know her because I’ve been her. This time last year, we were on our annual family camping trip and it hit me. I was scrolling through my phone organizing forms and to-do lists in my notes app. I was missing the moments right in front of me. I immediately put my phone down and joined my kids in the campground silent disco event. And let me tell you, my kids still talk about that evening. And to think I almost missed it.

I’ve written before about crouching right in front of my son, mid-conversation, while my brain was three aisles deep in a grocery store. August is that, industrialized. A whole month of being the last-summer-weeks mom on the outside and the September project manager on the inside.

And here’s the part I want to say carefully, because the internet is full of people who will sell you the wrong conclusion.

The answer is not to relax harder. You cannot breathe your way out of a real workload. The signup really does open at 8am. The forms are really due. Every “just be present” post is written as if the September project will staff itself, and you know exactly who it staffs itself with.

The answer is also not to hustle the list down to zero in July so you’ve “earned” August. I tried that. The list refills. It’s a tide, a season with its own weather.

The answer, the only one I’ve found, is boring: the September work has to live somewhere that isn’t your head. Not gone. Moved. The signup date on a system that will surface it. The forms in a place that reads them for you. The first-week logistics written down once instead of rehearsed nightly. None of that makes the work disappear. It makes the work stop broadcasting, all day, on the channel you were trying to watch your kids on.

Because that’s what the split really is. Interference. Two signals on one channel, and the louder one wins, and the louder one is always the one with a deadline.

The pool has no deadline. That’s the whole problem. The best parts of a life never do.

So this weekend, when the school PDF lands, and it will land, notice which version of you picks it up. If it’s the desk-in-her-head one, at the pool, on a Saturday: that’s the moment. Not a failure. A signal that a piece of September is still living in you instead of in a system.

The last weeks of summer are the part nobody gets back. They deserve better than being fall’s staging area.

If you know the mom who’s at the pool and at the desk at the same time, send her this. Not as a hint about her phone habits. As a hand on the shoulder. She’ll know which paragraph is hers.

Subscribe now

Leave a comment

Emergency contact form. Media release. Technology agreement. Allergy disclosure. The one where you write the same four phone numbers you wrote on the previous three. Handwriting, in 2026, information the school has had since kindergarten.

And that’s one packet. A school year is ten months of packets. Newsletters with buried signups. Spirit-week schedules that matter for exactly five days, but you’re shunned if you forget. Picture-day flyers. Field-trip permissions with deadlines and dollar amounts. The PDF drip does not stop until June.

Here’s what I finally admitted: I was the document pipeline. Every piece of school paper passed through my eyes, got sorted by my brain, and landed on my calendar by my thumbs. The reading was the job. Nobody counted it, including me.

You cannot delegate that job to a person. Your partner reading every PDF too doesn’t halve the work, it doubles the reading. This is the rare household job that is actually shaped like a machine’s job: unstructured paper in, structured lists out, every time, no judgment calls.

So during install week I built the household a paper-eater. It is plain on purpose: one standing instruction living in an AI chat that I feed every school document the moment it arrives. Photo of the paper form, forward of the email, paste of the newsletter. It hands back the same three things every time: what goes on the calendar, what needs buying, what needs signing. Then those go where they belong and the paper is dead to me.

It has run since June (summer camp life). It caught the days each of my kids needed to bring $3 for ice cream, which days they needed life jackets for the lake, the day they needed to dress like tacky tourists, when bathing suits and water shoes are needed, and much more. My part takes about two minutes per document, most of which is taking the photo.

Below the wall: the standing instruction verbatim, the twenty-minute install walkthrough, where the three lists land so the whole family can see them, what broke the first month and the fix, and the starter version you can run tonight before the next PDF lands.

Read more

It shows up between two work meetings. A PDF from the school: 24 pre-sharpened pencils, glue sticks but only the purple kind, a red plastic two-pocket folder with prongs, headphones in a labeled Ziploc. And further down, quieter, the line that actually matters: after school program signup opens September 1 at 8:00am. First come, first served.

Back-to-school doesn’t arrive as one task. It arrives as forty small ones wearing a trench coat, spread across four weeks, each with its own tiny deadline. The supply list. The beginning-of-year packet with eleven forms in it. The bus form. The sports physical that needs an appointment that needed booking in June. Picture day, which is a Thursday. One of the Thursdays.

At work, this would be called a project. It has a scope, a deadline that cannot move, and stakeholders who cry when it slips. Projects get project managers. This one got you. Nobody hired you, nobody wrote the job description, and nothing about the setup admits it’s a job. The school sends the list and assumes. The signup page opens at 8am on a workday and assumes. Every piece of it quietly assumes that each household ships this project on time, for free, every August.

I’m a civil engineer. I run projects for a living, and I can tell you why projects actually fail. It’s almost never effort. Teams don’t blow deadlines because they’re lazy. They blow them because forty tasks with forty tiny deadlines never got sequenced, so everything became urgent on the same Tuesday.

And the heaviest part of this particular project isn’t doing the forty tasks. It’s being the only one who knows they exist. The aftercare signup doesn’t send you a calendar invite. It’s one line in a PDF, and if you don’t catch it, there is no aftercare in the fall. Catching it is a job. Nobody counts that job.

So this year, run August the way you’d run it at work. Four weeks, four jobs, in order.

Week one is install week. Give the household a memory that isn’t you. Twenty minutes, one evening. That’s its own story, and I’ll tell it separately.

Week two is supplies and forms. The day the lists land: one consolidated cart, before the purple glue sticks sell out. An answer sheet built for every signup, so the 8am registration stops being a sprint.

Week three is systems. The school email pile crests in the last two weeks of August, so a ten-minute sweep goes on the calendar twice a week. The beginning-of-year packet gets eaten in one sitting, the day it arrives.

Week four is the handoff: First-week logistics on one page, handed to every adult in the rotation. Partner, grandparent, sitter. The “wait, who’s getting Sam” call does not happen this year.

Then there’s a morning. Usually the first Tuesday. The form is signed, the bag is packed, the pickup is covered, and your head is quiet at 8:15am. Nobody will notice, because this whole job is invisible when it’s done right. Success looks like nothing happened. Do it anyway. Not for the noticing. For the quiet.

You don’t need more discipline in August. You need August pre-sorted.

I wrote the whole four-week plan down, with the routine that runs each week, and put it in the kit I built. It’s called The Household Mainframe, it’s $37, and the August Sprint is the last chapter: mainframemom.com/household-mainframe. If September has ever ruined a perfectly good August, that’s the door.

Tell me in the comments: which back-to-school task ambushes you every single year? I’m keeping a list. Obviously.

Leave a comment

Subscribe now

Not the supply list. The supply list is loud. Purple glue sticks, red folder with prongs, headphones in a labeled Ziploc. Annoying, but loud.

The expensive sentence is quiet. It sounds like this: “The after school program registration opens August 1 at 8:00 AM. Spots are limited.”

One line. No bold. No calendar invite. If you catch it, fall childcare exists. If you miss it, you find out in September, at work, mid-meeting.

Schools don’t send tasks. They send prose. Somewhere inside every newsletter, welcome packet, and supply PDF are four or five actual deadlines wearing paragraph costumes. And the job of finding them has a name nobody uses: you.

Here’s the two-minute version of quitting that job.

Open the newsletter. Save the pdf and load it into a new AI chat. Or select all (CTRL + A), copy (CTRL + C), and paste (CTRL + V) it into your AI. Then say the following:

That’s it. That’s the Minute.

The newsletter goes in as prose. It comes out as a list with dates on it, and the 8:00 AM line is at the top wearing a flag instead of a costume.

You were never bad at keeping up with the school emails. You were doing document review for free, at 9 PM, on a phone. Stop volunteering.

Tomorrow I’m sending the long version: why August is a project, who got staffed on it without being asked, and the four-week plan that pre-sorts the whole thing. Subscribe and it lands in your inbox. The robots will find the deadline in it for you.

Subscribe now