Klerix Sentinel · Compliance & Audit AI

An always-on auditor for every agent action your fleet takes.

Sentinel is a compliance-grade audit substrate and a continuous risk engine. It records every click, keystroke, prompt, and model decision, then evaluates them against policy in real time. Built for the customers your regulators actually inspect.

100%
actions captured
0
data leaves your region
ISO 42001
AI mgmt. system
SOC 2 II
in audit
What Sentinel does

Four pillars. Zero magic.

Immutable trace

Append-only ledger with content-addressed snapshots. Replay any workflow step-by-step months later, with screenshots, payloads, and verifier confidence.

Policy engine

Express allowlists, blocklists, two-person rules, budget caps, and PHI/PII handling in a typed policy DSL. Policies block actions at commit, not after.

Continuous evals

Sentinel auto-generates adversarial cases from your live traces and runs a regression net on every release. Drift detected before it ships.

Regulator-ready exports

Signed PDF audit packets, SOC 2 evidence bundles, HIPAA access logs. One click — already mapped to control frameworks.

Technical specs

Engineered to ship, not to demo.

FrameworksSOC 2, HIPAA, ISO 27001, ISO 42001, NIST AI RMF, EU AI Act high-risk.
RetentionConfigurable 1–7 years. Bring-your-own S3 / GCS / Azure Blob with KMS.
DetectionAnomalous action sequences, prompt-injection signatures, exfiltration patterns, model output drift.
IntegrationsSplunk, Datadog, Chronicle, Vanta, Drata, OneTrust.
Where it shines

Built for the verticals regulators inspect.

Healthcare

HIPAA access logs auto-generated for every PHI-touching workflow.

FS & insurance

Maker-checker enforcement on payments and claim approvals.

Public sector

FedRAMP-aligned evidence collection, region-pinned data residency.

AI governance

Continuous model-card updates and red-team eval history.

Ready when you are

See Sentinel run on a workflow from your stack.

A 60-day pilot on a single workflow. Hard SLOs. No procurement gymnastics.