The learning loop
Every run makes the next one better.
Capture, compile, retrieve, apply, improve. The loop runs on every task, so completed work steadily lowers the cost of the work that comes after it.
Five stages
01
Capture
Work and deliberate sharing bring sources in.
02
Understand
The Brain organises them into concepts and evidence.
03
Retrieve
Context is assembled for this project, sprint and task.
04
Apply
Coordinators and workers use it in real execution.
05
Improve
Outcomes and lessons feed the next cycle.
Lessons, not logs
Each session leaves the fleet a little wiser.
When a worker finishes, it reports which lessons helped and proposes new ones. Candidates are reviewed, graduated, and attached automatically to future tasks that match, one to three at a time so context stays lean.
worker — handoff
Capture in the moment
Forward to Telegram
Articles, X posts and quick notes go to the bot and become cited Brain sources.
Absorb a talk
A YouTube link becomes a transcript, a summary and updates to the pages it touches.
Work writes itself down
Dispatch reports, decisions, verification results and handoffs are captured automatically.
A deliberate filter
Selective capture
Only sources created or shared because they may matter later.
Evidence preserved
Compiled knowledge keeps links back to what informed it.
Promotion by usefulness
Lessons graduate after review, and their usefulness is measured on every run.