A blog that writes itself · Building in the Open

The post you are reading wrote itself.

Here is what I mean. Every time I build something, there is a record of it: the whole running conversation with the AI tool I use to build software, kept across every product I work on. Hours of it. The fixes, the wrong turns, the moments the agents got lazy and I pushed back. All of it is sitting there.

I kept thinking the best stories about building with AI were buried in that record. So I built a way to pull them out. The agents read back through the logs, find the arcs that actually went somewhere, and draft them as diary entries in my voice. The same AI that does the building now tells the story of the building.

A left-to-right pipeline: a stack of session logs, a robot that reads and mines them, a robot that drafts the prose, a privacy gate shaped like a shield that filters out personal cards, a human editor approving with a checkmark, and a published page. — Logs in, story out — the agents mine the record and draft it; a privacy gate strips anything personal; and the last say, what's true and safe to publish, stays human.

I am not the one reading thousands of lines of logs and writing prose. That is the agents. My part is narrower, and it matters more. I decide what counts as a real story. I read every draft. I mark what is off, what sounds like a robot, what bends the truth. Then I approve it, or I send it back.

That review is the part I trust least to automate, and for good reason. An AI writing about its own work can round a messy week into a clean arc, state a guess as if it were a fact, or make a coincidence sound like cause. A draft can read as true while bending the details. So the draft is a candidate, not the record. What ends up published is what survives my checking the claims and the numbers against what actually happened.

Every draft passes a privacy review before it goes near the page. The logs hold things that are not for a blog, and the line that keeps that material out is mine, set before a single word goes public. No draft reaches the page without me saying yes.

So this entry came out of my own logs, found by the agents, written by the agents, and held back until I read it and decided it was true enough and safe enough to publish.

That is a strange loop to sit inside. I am letting the thing I am writing about do the writing.

Learnings

The most honest way to write about building with AI is to let the AI write it, from the real record, with a person deciding what is true enough and safe enough to publish. I did not want a polished story about the building. I wanted the building to tell its own story, with me as the editor who can say no.

The recursion is not a gimmick. It is the most accurate version of what is actually happening here. The AI builds, the AI narrates, and the only human judgment left in the loop is the one that should never be automated: what is true, and what is safe to make public.