AI Co-pilot
An analyst co-pilot that summarises investigations, suggests next steps, and drafts response actions — with governed egress and a human always in the loop.
The Nirvet AI co-pilot accelerates the parts of an investigation that drain analyst time: building context, summarising timelines, and proposing response actions. It runs behind an egress-redaction fence and an admin-controlled provider allowlist, so sensitive data is handled on your terms — and every suggestion is a recommendation a human confirms, never an unattended action.
Investigation summaries
Turn a sprawling alert timeline into a concise narrative — what happened, which entities are involved, and what to check next.
Egress redaction fence
Outbound prompts are redacted against a zero-config floor before they ever reach a model. Empty config never means 'send everything'.
Admin-controlled providers
You choose the LLM provider and an explicit model allowlist per tenant. Governance is a configuration record, not a code change.
Evaluated prompts
Prompts live in a versioned registry with a hermetic evaluation harness and a CI fence — changes are tested, not shipped blind.
What you get
- Human-in-the-loop by default — the co-pilot proposes, an analyst decides
- Per-tenant data isolation carries into every AI interaction
- Redaction is applied at the boundary and audited
- Customer-approval and retention controls for AI-processed data
- No training on your data
Put the co-pilot to work in a live investigation
See how summarisation and suggested actions shorten triage — with the redaction fence and provider controls switched on.