The proof · my own brain, measured

Thirty-three days of growing a mind.

This is not a mockup of what CoBrain could do. It is what the system I run actually became, from a blank folder to the operation behind everything I make. Every number below was counted in the files, not estimated.

DAY 0 · JUN 10 2026DAY 33 · JUL 13 2026
The seed

It started as seventeen files that knew nothing.

Day zero was a starter kit: three template files, a single capture skill, and a promise that once the brain knew me a little, I would never have to over-explain again. Seventeen files. Zero memory. A dormant thing waiting to be woken up.

The measure

From nothing, in thirty-three days.

Brains1 blank3walled brains + 4 ops layers
Wiki articles0194
Words distilled0171,000
Raw sources captured0199
Outputs generated04,075
Lessons it wrote about how I work0338
Agents on the team019
Skills, live112
Prompt systems018
Total files174,817

Counted July 13, 2026. The system has kept growing since; these are the last numbers I sat down and verified file by file.

The architecture

It grew layers that didn't exist at setup.

The kit shipped with one brain. What runs now is an operating system.

Shipped in the kit

The starter brain

One blank brain and the capture pattern. The living cell everything else grew out of.

captureraw → wiki → outputs
Grew: knowledge

Walled worlds

Separate brains for life and each business, kept from bleeding into each other on purpose. The walls are the point.

3 brainswalled
Grew: a team

Agents & prompt systems

Nineteen specialized agents and eighteen reusable systems: a front-line audience agent, a research fan-out, a writing system, an accountability coach.

19 agents18 promptsself-learning
Grew: senses

Live integrations

It stopped waiting to be fed. It reaches into a dozen tools and pulls the world in on its own.

emailNotioncalendarmeeting recordersDriveWordPressKit
Grew: a nervous system

Routines that run while I sleep

A live command center wired to autonomous jobs that sweep new recordings, file them into the right brain, rebuild the morning briefing, and track every open loop. Every morning at 5am, before I'm awake.

5am sweepsmorning briefingopen loops tracked
The real story

It didn't just get bigger. It got smarter.

Size is easy to fake. This is the part that isn't: the brain learns from its own mistakes, teaches the other agents, and never needs the same correction twice.

01 / it remembers how you work

338 lessons, written by the brain about me.

Every session, a habit fires: if it got something wrong or watched me rewrite a draft, it saves the lesson so the next first draft starts closer to my final. Not settings someone typed in. Corrections it made to itself.

learned: no em dashes, anywhere. learned: read the whole thread before replying. learned: surface names, never database IDs. learned: soul over polish.
02 / it teaches the other agents

A lesson taught once lands everywhere.

When I correct one agent, the fix doesn't stay in one chat. It gets written into that agent's shared rulebook, so the rule is live the next time it wakes, in every conversation.

03 / it works unprompted

The morning is already handled.

New recordings filed, the briefing rebuilt, the open loops re-checked, all before 6am. I read; it did the remembering.

Now packaged

CoBrain is this exact system, generalized so you can install it and grow your own.

Join the waitlist