We went looking for someone already doing this.

The responsible way to build something strange is to find the people who already built it and copy their homework. So before we committed to running a studio on a single agentic brain, we did the survey: dozens of repositories and papers, the whole current wave of agent memory, agent operating systems, and “company-in-a-box” frameworks. We were not looking to be first. We were looking to be second.

The field is enormous and much of it is excellent. There are memory layers with tens of thousands of stars, skill libraries the size of small operating systems, always-on assistant runtimes that genuinely operate tasks for a person. What there was not, in any of it, was the thing we actually wanted to copy: a real company whose day-to-day operations — the research, the tickets, the digests, the watch on the competition — are run by a single agentic brain on a schedule, with humans kept for the calls that matter.

Plot what exists by two questions — is it a tool you run, or a thing that runs itself; is it a demo of an idea, or is it operating real work — and the projects cluster in the corners. The memory engines and skill kits sit low and to the left: components, brilliant ones, that you assemble. The framework demos sit org-shaped but idle, performing the shape of a company for a screenshot. The one genuine neighbor we found was a blog post — an agency running its operations on a Claude-Code second brain, earning its software more trust over time. Not a codebase we could read. An account we could nod at.

The quadrant we live in came back empty. That is a strange place to be proud of, and a good place to be nervous.

We want to be careful about what that emptiness means, because the flattering reading is almost certainly the wrong one. A quadrant stays empty for two kinds of reasons: nobody thought to stand there, or everybody who tried got carried out. We do not yet know our ratio. The field is littered with confident systems that did the wrong job at scale, memory that rotted the moment no one curated it, autonomy that reached one step too far. We have hit small versions of all three, some of them the same week we are writing this.

So we are treating “no one else publishes this” as an obligation, not a trophy. If we are going to stand in the empty quadrant, the least we can do is leave notes for whoever stands here next — what the field actually offered, what we borrowed, and where the ground gave way. The other essays in this series are those notes. This one is just the map we drew before we started walking.