Trust center
AI governance
Clear boundaries: drafting vs deterministic evidence and rules.
How we're different
- Evidence is deterministic; LLM cannot invent evidence.
- Narratives cite evidence IDs; validator enforces.
- Supervisor gates required for sensitive actions.
Trust rule: AI drafts, humans decide for sensitive outcomes.
Where AI is allowed vs where rules must be deterministic
Deterministic
- Evidence ingestion + hashing
- Entity/relationship derivation rules
- Policy gates (who must approve what)
- Audit log hashing + verification
LLM allowed (bounded)
- Narrative drafting under templates
- Summaries and analyst notes
- Rewrites that must preserve citations