Trust center
Privacy
Data minimization and clear boundaries for model usage.
How we're different
- Evidence objects with provenance and hashes.
- Narratives cite evidence IDs; QA validates citations.
- Exports are audit-first and verifiable.
Trust rule: AI drafts, humans decide for sensitive outcomes.
Data handling
Data minimization
Store only what is needed to operate investigations and prove evidence for audits.
- Evidence objects (with hashes + provenance)
- Narrative drafts and QA results
- Audit events and export proofs
Retention and deletion
Retention is policy-controlled and environment dependent (roadmap for enterprise controls).
- Per-org retention defaults
- Export before delete workflows
- Legal hold (roadmap)
Model usage boundaries
- LLMs may draft narrative language and summaries.
- LLMs are constrained to cite evidence IDs; QA rejects uncited claims.
- No training on customer data unless explicitly agreed (policy stance for planning).
Not legal advice. Requirements vary by jurisdiction and institution.