Review answer quality after deployment.

Scheduled review of captured answers or generated probes

Watchtower evaluates answers after delivery; it does not block production actions. Captured conversations reflect observed use, while endpoint probes are generated tests. The record distinguishes their scope.

• Passive JSONL or HTTP capture and active endpoint probing
• Hourly, daily, weekly, monthly, or manual cadence
• Per-cycle sample controls and recorded exclusions
• Session context where supplied with captured conversations
• Retained findings and recurring failure patterns

Evidence you can retain, share, and inspect independently

Successful, seal-eligible paid audits produce readable reports and machine-readable findings, with recorded methods, scope, and limitations. A signed manifest and portable checker let recipients detect altered, missing, or added files without a VeriTrooper account or installation. Human signoff is included when recorded. Evaluation provides a limited proof set; companion documents depend on the selected checks.

• Bound Monitoring Report and Executive Summary
• System Card, Monitoring Scope, and Monitored Traffic Card
• Flagged answers, supporting evidence, and review dispositions
• Gap and Remediation Diagnostic plus Data Findings
• Framework Crosswalk and optional Annex IV Monitoring Record
• Evidence JSON, model-card YAML, SARIF findings, and custom evaluation records
• Recorded configuration, cycle records, signed manifest, and portable checker

Inside the approved boundary

Capture and scheduled analysis can run locally. Remote endpoints are contacted when selected for probing or model review. Captured turns are evaluated during a scheduled or manually started cycle, not during the capture request.

Use the findings in your existing review process

Custom evaluation-record exports may require an adapter for your tracing system. Examine the recorded sample, unresolved cases, and exclusions before drawing conclusions about wider production traffic.