Architecture · arrives in Phase 3
AI Architecture
The AI-native system diagram: user → inputs (manual logs, wearables, cycle history, symptoms, workouts) → data layer → five domain engines → insight layer → recommendation layer → UI cards → user actions. Intelligence emerges from multiple systems reading one record — not from a single score.
Scheduled for documentation Phase 3
This section's structure is reserved and its source material already exists in the shipped application. It will be populated when Phase 3 of the documentation package is delivered — see the roadmap.
What this section will contain
- —The full system diagram from input to action loop
- —Engine composition: recovery, cycle, training, nutrition, symptom engines over the shared Entry record
- —Insight layer: pattern detection, confidence, the Pattern→Insight promotion path
- —Recommendation layer: how engines' outputs reconcile (symptoms always win)
- —Explainability plumbing: signalsUsed, assumptions, missingInputs from engine to card
- —Why no single score: the case for capability bands + domain playbooks
Grounded in (already shipped)
- ·The ten AI Principles (already live in this portal)
- ·Engines run once per tab render; results passed down — no duplicated inference
- ·Cross-signal priority audit (Phase 6 of the product) and wearable integration (Phase 7A)