What is published here
This site holds conceptual documentation: how the trail is made, what a record contains, and how a CSV export is read. It is not a full product manual and not a public API reference.
- Control loop: observe, detect, trace, scope, prove.
- Export format: column-level CSV reference for audit-log-entries.
- Hash chain verification: how a reviewer checks entry_hash / prev_hash.
- Security model: customer-owned environment (see Security).
Doc index
Start here when you need something specific:
- Export format (/docs/export-format): CSV columns, empty fields.
- Hash chain verification (/docs/hash-chain): walking entry_hash and prev_hash.
- Product (/product): audit log, record detail, policy on the row, live trail.
- Security (/security): hermetic, customer-owned boundary.
- What is an AI audit trail (/what-is/ai-audit-trail) vs observability and DLP.
What belongs in docs later
When the product is ready to document beyond marketing, add pages in this order, still conceptual where possible, deep runbooks only inside the customer environment:
- Record lifecycle: create, open, export, retain, redact.
- Policies on the row: how policy IDs are assigned, including shadow AI.
- Toolbox coverage: which AI systems can appear on source_type / ai_system.
- Roles and access: who can see the live trail vs who can export.
- Retention and legal hold: shaped per enterprise (no generic SaaS defaults claimed here).
Skip public connector SDKs, ingest APIs, and deployment runbooks on this site until they are real and supportable. Those travel with the customer-owned environment.
Control loop
Every surface is a view on the same loop: observe AI use, detect untracked or policy-relevant events, trace who and which system, scope to the environment and policies you define, prove with an exportable trail.
Related