System map

This map describes the product without exposing private infrastructure.

You or a schedule
        ↓
Project context and current evidence
        ↓
Bounded plan with limits and checks
        ↓
Local AI worker in an isolated workspace
        ↓
Tests, real-target review, and human decisions
        ↓
Result, recovery path, and updated project memory

Direction layer

Briefing, Projects, Schedules, and Ask North Star help you choose or understand the next useful work. They all resolve to the same project-aware task model.

Execution layer

The server prepares a worker with the required context, tool access, and limits. Work can start immediately or through a schedule. Resource limits and protected actions remain in force.

Review layer

North Star collects automated checks and inspects the real target when the task requires it. A protected decision returns to the operator with evidence.

Memory layer

History stores the result. The project current state changes only after verification. The next run begins with that compact state and can follow references to deeper evidence when needed.

Public layer

The public preview, case study, and these docs are static explanations. They do not connect to the execution layer or expose the protected dashboard.

Next: Plain language glossary