AI Assistant

Auditing AI Activity, Latency & Query Analytics

Monitor customer interactions with your AI assistant, review query confidence scores, inspect source tokens, and track response latency in milliseconds.

Why Audit AI Activity?

Auditing AI activity logs allows business managers to see what questions customers are asking, verify that the AI is retrieving the correct Knowledge Base policies, and identify missing product information.

Viewing AI Activity Logs

Navigate to Dashboard -> Settings -> AI Activity Logs (`/dashboard/ai-activity`):

Customer Query Inspection: View the exact customer prompt and the generated AI response.
Source Data Breakdown: Inspect which Knowledge Base entries or catalog items were retrieved and used as grounding context.
Confidence & Latency Metrics: Review response generation latency (in milliseconds) and model confidence indicators.
Intent & Outcome Tags: Filter logs by intent ("buy", "browse", "support") or outcome ("resolved", "escalated").

Using Log Insights to Improve Grounding

If logs reveal repeated queries where the AI replied with low confidence, add a dedicated entry to your Knowledge Base covering that topic to improve future automated answers.