The company I wanted to work in

Nobody was ever going to say it

There was no map of the project. There was no document explaining why anything had been built the way it was. The knowledge lived inside one person who had started it and still held it, and even where a structure was supposed to exist, that person changed things without telling anybody.

Both of those companies were badly run, and I am not going to be polite about it any more.

Everybody knew. They talked about it constantly, among themselves, in the way people talk when complaining has quietly become the substitute for changing anything. And nobody ever did anything.

Once, I did. I tried to fix one of the things everyone had been complaining about for years.

Every single person who had agreed with me in private went quiet.

That is why I do not work there any more.

So I am building the opposite, before there is anyone to hand it to

I am a solo founder. I manage my own health, I write the application, I run the company, I keep the accountant's folder, and I am building the infrastructure for other people to take over.

That last one sounds strange, because there is nobody to take it over yet. That is exactly why it is being done now. The person I eventually hire is going to inherit the thing I was never once handed.

There is a folder of decisions with the reasoning kept, including the wrong turns. There is a runbook for the migration. There is a log of what happened on which day. The accountant already has a folder with everything in it. None of that is impressive on its own. What makes it worth doing is that it exists before it is needed, rather than being reconstructed afterwards by somebody guessing.

Everything is written down, not because documentation is virtuous, but because I know precisely what it feels like to arrive somewhere and find nothing.

Where the product came from

It was not a product. It was a set of governance documents I wrote to tell a language model how to read my own blood work, wired up with APIs and JSON, for an audience of one.

I built it because I wanted to understand what was in my own results rather than being handed a conclusion. Then I saw my scans. I saw cancer free. And I saw how much changed once I actually understood what was happening in my body, because understanding it is what let me ask better questions.

That is the moment it stopped being a personal tool. Not a market study, not a gap analysis. Just the realisation that the thing sitting on my laptop was the thing I had needed all along.

What the infrastructure costs

In four days I took two sites off a builder and rebuilt them on my own infrastructure. Two days each. Both are live.

Hosting the company site costs about 45 cents a month.

That is not a prototype, and it is not a free tier trick that expires. It is object storage, a content delivery network, DNS, a certificate, a deployment pipeline that holds no credentials at all, and gates that refuse to publish when something is wrong. It runs on the same platform large companies run on.

I want that number to land, because the thing standing between most people and owning their own infrastructure is not money. It is the belief that this class of thing is out of reach, and it is not.

What a builder is actually good for

I do not want this read as a complaint about builders, because I got real value from one.

A builder is a development system. It is very good at taking an idea that exists only in your head and giving it a shape you can look at, quickly, before you have committed to anything. For someone who does not write code, that is the difference between an idea and nothing at all.

What you cannot get out of it is the architecture. You can watch the thing working and still have no picture of how it is put together, and no way to take that picture with you.

So if you are a programmer, you can have both, and most people in that position do not realise it. Use it for the shape. Then take the shape somewhere you own, and build the structure yourself.

And the assistant, openly

I decided some time ago to stop keeping everything in my head. The ideas are mine, the architecture is mine, the decisions are mine. What changed is that the structure I could already see now has somewhere to live outside me, and I am not carrying all of it alone.

It runs through the whole day rather than only the coding hours. Calories. Exercise. My own results. The application. The company.

I am not planning to work any other way again, and I would rather say that plainly than pretend otherwise in public while doing it in private.

It is also not free of judgement. It broke my deploy gate and two builds failed on a site that was perfectly fine. Its notes about my certificates were wrong in a way that would have taken my company site down had I followed them without looking. It created a database in the wrong country. Every one of those was caught because I checked, not because I was told.

So the choice that actually matters is not whether to work this way. It is which model and which assistant you choose, and how carefully you read what comes back.

What I am actually building

A company where the thing in my head and the thing on the disk are the same thing.

A place somebody will one day be glad to arrive at, because on their first day they will be handed the map instead of the chair.

And when I say I am building the company I wanted to work in, I do not only mean the documentation. Documentation is the cheap part. I mean a place where the person who points at the problem is not left standing there on their own.