Platform · Observability & audit

Evidence for every governed call.

IntSen keeps request and usage records, policy verdicts, turn records and evaluation results in one place—with operational metrics, logs and alerts to run it with confidence.

Book a live demo View the quickstart

One request record

request_id: req_01JD8M2Q4N
registration: field-service-assistant-prod
agent: field-service/fleet-service-advisor
client (verified): fleet-advisor-prod
user: sha256:9c1e…(hash only)
model: gpt-5.4-mini
verdicts: pii: masked · injection: pass
cost: $0.0031 · 1,004 tokens
status: 200 · 912 ms
body: stored 90 days · PII redacted

Illustrative product scenario — synthetic data

The problem

When something goes wrong, the evidence is everywhere—or nowhere.

Investigations stall when prompts weren't kept, logs are spread across services and nobody can link a complaint to a specific call and decision.

No single record of calls, decisions and outcomes
Prompts and responses either kept forever or not at all
No alert when usage looks unusual

Capabilities

What Observability & audit does

Request records

Caller, use case, agent, model, tokens, cost, timings, status and refusal code for every governed request.

Policy verdicts

Every guardrail decision, with category and score, linked to its request and turn.

Task evidence

Turns with their steps, scores and compliance findings.

Body capture policy

Store request and response bodies per use case, for your retention period, with personal-data redaction.

Operational metrics

Volume, errors, latency and verdicts by team, use case, client and model; logs for debugging.

Alerts

Budget, quality, compliance and usage alerts to owners; operational alerts to the platform team.

Technical detail

Where the evidence lives

Each IntSen service writes its own records to a shared database; events flow between services through an outbox.

Request records: database · per request · cost, timings, status, refusal code
Verdicts: database · per check · category and score, never the matched content
Turns and steps: database · per turn · steps, totals, scores
Bodies: object storage · only if capture is on · retention days · PII redaction
Metrics and logs: metrics, logs and dashboards for operations

Why it matters

For the people accountable for it

CISO / risk

Investigate a governed turn from complaint to call to decision.

AI platform

Operate the platform with metrics and alerts per team and use case.

Procurement / legal

Capture and retention set per use case, matched to its data class.

Boundaries

What to know before you evaluate

We would rather you hear the limits from us than discover them in a pilot.

Records cover governed traffic only.
An auditor-ready evidence export and governance control checks are planned for after the first release.
Bring-your-own storage for request bodies is planned; today bodies are stored in IntSen-managed object storage.

FAQ

Common questions

AI Gateway Agent tracing Guardrails

How long are bodies kept?

For the retention period set in the use case's capture policy. Body capture can be switched off entirely.

Is end-user identity stored?

Only as a hash of the user id your app sends.

Can owners read captured bodies?

Yes, owners can read captured bodies for their own use cases in the Console.

Ready to make your enterprise AI accountable?

See how IntSen can help your teams govern access, safety, quality and spending across the AI applications and agents you already run.

Book a live demo Discuss an enterprise pilot