Nobody pays a personal trainer for their knowledge. They pay for the momentum.

In February 2026 I wrote this in a paper notebook, about going to the gym:

Starting a habit is hard, the key is about building the momentum. If you already get the wheel going, you are already 50% successful. But at the beginning, the wheel will definitely move very slowly. It will demotivate your energy. "Why am I doing it so slow? It's so hard!" you are absolutely right. That's why you will need someone to hold you accountable for. A PT in doing gym, a teacher in learning piano — you name it! They aren't paid for their knowledge — we can learn everything online now. What we are paying them is the responsibility to keep the momentum.

I was writing about exercise. I was describing my job.

The knowledge is free. It has been free for years.

Every technical thing I do for a client is documented publicly, for free, by someone better than me. The database, the deployment, the integration — all of it is a search away, and now an AI away.

If knowledge were the product, my work would be worthless. So would a personal trainer's. So would a piano teacher's. Nobody needs a human to find out that squats exist.

What people can't get from a search result is the thing that happens on the eleventh day, when the wheel is still moving slowly and the obvious conclusion is this isn't working.

The flat part is where projects die

Every build has the same shape. Week one is fast, because everything you touch is new and every commit visibly changes something. Then comes the part where the work is real: edge cases, the data that isn't shaped like the demo, the integration that half-works.

Output looks flat. Effort doesn't drop — it goes up.

That is exactly the moment a project gets quietly abandoned, or rewritten from scratch by someone who thinks the flat part is a sign the approach was wrong. Usually it isn't. The flat part is just what the middle of a build looks like from inside it.

Somebody has to be responsible for the wheel continuing to turn through that stretch. That responsibility is the product.

What that means in practice

Three things I do that have nothing to do with technical knowledge, and everything to do with momentum:

Ship something visible before the hard part. Not a mockup — something the client can open on their phone and show to a colleague. It buys the goodwill you will spend during the flat weeks.

Make progress legible when it isn't visible. During the flat part I report what got harder and why, not just what got done. A client who understands why week three looks slow does not panic in week three. A client who is only shown a burndown chart does.

Keep the smallest possible daily step alive. From the same notebook, two years earlier, on why homework habits kept failing me: "after seeing that I have to pay lots of effort right at the beginning of the course, rather than a few step everyday, discourage me." A build that requires a heroic week is a build that will stall. A build that requires a boring daily step usually finishes.

Today just 1%

The end of that February note:

Slow but it doesn't mean you are not making progress! Today just 1%, tomorrow still 1%, but then 5%, then 10%, then you don't even know how fast you are going.

I have been building software with AI for months now, on real client systems that real shops depend on. The technical parts I got wrong, I got wrong loudly and fixed. The projects that went badly were never the ones with the hardest problems.

They were the ones where the wheel stopped, and nobody was responsible for starting it again.

This is from a notebook I have kept by hand since 2024 — 119 pages, transcribed word for word. This is post one of five.