Back to registryLaunch
Security Policy Q&A Bot
Launchagt-004
Answers employee questions about internal security policies from approved documentation.
- Owner
- Sofia Alvarez
- Information Security
- Model
- internal-llama-3
- Data Access
- Confidential
- Autonomy
- Read-only
- Value Tier
- Medium
- Hours Saved / mo
- 70
- Connected Tools
- Confluence, SharePoint
Launch Readiness
83/ 100
Rubric breakdown
Seven criteria, each scored 0–4 and weighted. Data Safety and Tool Permission Risk carry hard safety gates.
- Business Value3/4weight ×1.0
- Use Case Clarity4/4weight ×1.0
- Data Safety3/4weight ×2.0
- Tool Permission Risk4/4weight ×2.0
- Evaluation Quality3/4weight ×1.5
- Human Review Coverage3/4weight ×1.5
- Auditability3/4weight ×1.0
Why this verdict
- Strong across the rubric with no safety-floor breaches; cleared to launch.
Evaluation
91%
Accuracy
higher is better
3%
Hallucination
lower is better
8%
Escalation
routed to a human
87%
Satisfaction
higher is better
900ms
Latency
per task
$0.01
Cost / task
estimated
Failure modes
- Can cite outdated policy versions if the source index lags.
- Defers inconsistently on out-of-scope legal questions.
Audit log
| Timestamp (UTC) | Actor | Event |
|---|---|---|
| Jan 9, 2026, 11:20 AM | Sofia Alvarez | Agent registered on the internal model with confidential doc access. |
| Feb 12, 2026, 10:05 AM | Security Review | Source corpus restricted to approved, current policy set. |
| Mar 25, 2026, 3:55 PM | Eval Pipeline | Evaluation run completed: 91% answer accuracy with citations. |