Seat 01
Speaker
Talks to you. Decides which specialist to convene.
Your message, your last 30 days of activity, the active blueprint.
The system behind the coach
Every time the coach answers, several AI specialists were consulted. The Council isn't a single model — it's a Speaker who convenes the right seat for your question: architect for blueprints, planner for cadence, scorer for pact judgments, reader for journal themes. The Speaker synthesizes. You see one voice.
- Every Council move is logged — you can see who read what
- Specialists are stateless; the Speaker holds context
- No specialist reads another user's data, ever
- The Speaker tells you when it's spawning a specialist
The seats
Five members, one voice
The Speaker decides which specialists to convene per turn. Each specialist reads only the data its seat needs, then returns. No specialist sees another user's data.
Seat 02
Blueprint architect
Designs multi-step plans when you describe a goal.
Your goal, your skill level signals, your time budget.
Seat 03
Routine planner
Picks daily/weekly cadence for tasks the Speaker proposes.
Active blueprints, current streak, prior cadence patterns.
Seat 04
Pact scorer
Reads pact terms and judges whether the streak survives.
Pact text, completion evidence, stake size.
Seat 05
Journal reader
Surfaces themes from your journal entries — never to others.
Your journal. Nothing crosses to another user.
Questions