# Advisory AI

## Purpose
Evidence-grounded analysis with guardrails and offline outputs.

## Inputs
- SBOMs and evidence bundles

## Outputs
- Structured findings and guidance artifacts

## Data and storage
- PostgreSQL and artifact store

## Key dependencies
- Scanner outputs
- Policy evidence

## Notes and boundaries
- Guardrails required for outputs

## Related docs
- docs/modules/advisory-ai/architecture.md