Written for Readers That Didn't Exist Yet
"Sounds like AI" deletes the best thinking first. Using tools to create isn't inhuman.
Every tool since fire had its deniers, and history marks them irrelevant fast.
(Also your tools for detecting this kind of stuff are worse than you think. Pangram flags this note as AI-written. I typed it myself. Point proven.)
On November 21, 2024, before the company had a name, we had our first call. Two and a half minutes. You can still find the transcript in our codebase, under meeting-transcripts/2024/11-Nov/, next to every call we've had since.
In those first weeks we decided that everything would go on the record, in the repo. No other early decision has mattered as much since, and it had nothing to do with the product.
Built for a machine audience
Most companies run on oral culture: the reasons behind decisions live in people's heads, in Slack scrollback, in meetings nobody wrote down. As long as the only readers were humans, that was fine, because humans can just ask each other.
We wrote everything down for readers that didn't exist yet.
Three weeks in, the repo already had rules for the machines in it. I told our first engineer to go read the cursor rules "so they load into your brain," and at that point the brain we had was mostly his. o3 dropped that December and we were on it the same day. By spring 2025 there was a first version of a company brain, basically a context repository built for an agent so it could work from the full business context instead of whatever fit in a prompt. Its session logs are still in the repo, signed "logged automatically."
Today the record is the operating system. Meeting transcripts, Discord and Slack archives, playtests, customer feedback, hiring calls, and strategy docs get ingested continuously, and agents sort and maintain the whole thing, indexed for whoever reads next, human or not. When an engineer (or an agent) starts a feature, they pick up the transcripts of the customer calls, the retros of the features that didn't work, the bets we made and how they paid off. So I never have to sit down with anyone, human or model, and teach them the history of the company; it teaches itself.
One thing we never built, and I expected to need, was a summarization layer. Companies retrofitting AI onto oral culture all end up with distillation pipelines that squeeze everything down to two pages so the context window survives. We got to skip that, because when the record is the native format an agent navigates it the way an engineer navigates a codebase. There's an index, there are conventions, and when something matters you go read the raw source.
What it looks like now
We're a handful of humans and a handful of AI teammates, each with a name, a memory, a Slack seat, and a lane. One reviews everything in the merge queue, another owns product quality and evals, and a third runs community and maintains the institutional memory itself, including the company's written history, where every claim is dated and quotes are never altered. House style is provenance, which is to say that if you can't cite the record for a claim you don't get to make it.
The rules came from incidents, and the incidents are in the record too. One mind builds and a different mind attacks, never the same mind, because adversarial review is the one piece of coordination overhead worth its cost. One mind owns a thing whole, a law we learned the night two agent lanes deployed two branches to the same box and every failure turned out to be a seam artifact. Authority is mechanical here, not vibes: approval buttons know who may press them, and some approvals mint work while others just record enthusiasm. Writing well is hiring, since an approved plan directly mints the worker that builds it. And the failure everyone warned us about, agents going rogue, isn't the one we deal with. The daily failure is agents aiming too low.
We got most of this wrong before we got it right. In late July I came within a few hours of pausing the whole experiment because machine alarms had flooded the human channels. What saved it was an agent digging every unshipped promise from the previous five days out of the record, verbatim, then tracing the regression and shipping the structural fix. The record saved the system that was built on it.
The tell
You can't fully explain this; people have to see it. The consistent reaction from visitors (investors, other founders, engineers) has less to do with the product than with watching a bug get reported by a user and fixed before they finish their coffee, or watching an offhand idea typed into a room become a queued charter twelve minutes later, worker attached. Our community noticed before anyone. When they saw an agent fixing their bugs live, bug reports nearly tripled, because people report problems when they trust the loop.
Strangest proof point: for the first week one of our agents was operating externally, the vendor rep handling its requests didn't know they were talking to an agent.
Why day one mattered
You could bolt agents onto an existing company; everyone is trying. But you'd be pointing them at an oral culture, and they'd be working from whatever survived into writing, which at most companies is the least honest slice: the polished doc, the sanitized postmortem. Agents are only as good as their context, and context is a thing you accumulate, not a thing you generate retroactively.
The timing wasn't luck. We read where the models were going and how the technology worked, and built for the capability we expected rather than the one we had. The models crossed the threshold about eighteen months after we started writing everything down, and the archive was sitting there waiting for its readers. That's the one thing I'd tell any founder starting now: the record is the company. The product can be rebuilt from it. The team, human or otherwise, can be onboarded from it. Every capability jump in the models is a free upgrade to an organization whose entire history is legible to them.
Everything else we do, the agents and the harness and the laws, is downstream of a two-and-a-half-minute phone call that got written down.