Four things you actually control

The AI divide from the last story can feel huge and out of your hands. It is not. When I worked out what a solo builder can actually own, not rent, I landed on four things. Here they are, and why they only really work together.

One: expert playbooks

The hard part is learning. So the first thing you own is the learning itself. Not vague advice. Real playbooks. Books, field notes, and honest build stories that turn hard lessons into steps you can follow.

This is why I write the way I do, and why this whole story log and the books exist. Watch someone do the real thing, with the real numbers, and suddenly you can do it too.

Two: open software

Next you own the tools. Open software means you can run the AI helpers yourself, for free, and look right inside them. No per-seat fee. No cap on how much you automate. No vendor who can change the deal on you next quarter.

That is the software I build. Yours to run. Yours to change.

Three: custom models

Then you own the brain. Not one giant chatbot that knows a little about everything. A custom model, taught one field, like patents or security or law or money or health. Smaller, sharper, and tuned to your actual work.

That is the models work, like Patent Strategist. Open weights, your field, your machine. Story ten was me building exactly this kind of model on my desk.

Four: a personal device

Last, you own the ground it all runs on. A personal device, a computer strong enough to run real AI privately, on your own desk. Story nine was my first one, a small machine the size of a book.

When the brain sits on hardware you own, your data never has to leave the room. Privacy stops being a setting you switch on and becomes simply the default.

Why four, and why together

Notice the through-line. Every one of these is something you own, not something you rent. Renting means locked in, metered, and gone the day the terms change. Owning means private, yours, and growing in value over time.

And they feed each other, like a wheel that keeps itself turning. Playbooks teach you the craft. Software does the work. Custom models make the work yours. A device keeps it all private and cheap to run. Pull one lever, and the others get easier.

Three things make right now the moment for this. New tools every week, so owning your learning beats any one-time plan. Skill beats staff, so one person with AI helpers can do what used to take a team. And open tools have caught up, so free now matches paid.

Where to start

You do not have to grab all four at once. Pick one lever and pull it. Read a playbook. Run one open tool. Try a small model. Get a machine of your own.

The rest of this log is me pulling all four, out in the open. The next one looks up close at the software, and the small kit that makes building it far less painful: The glue tax, and the kit that pays it.