AI Website Chat Agent
Runtime evidence

AI Chat Logs

Inspect AI input, answer previews, runtime, model, response IDs, and handoff flags for each chat.

Review AI logsInspect runtime evidence for generated answers without exposing provider secrets.

AI performance review loop

Governed review classifies visitor intent, answer action, relevance, and one root cause before proposing any draft. AI reviewer flags unanswered, weak, risky, off-brand, handoff-friction, and missing-contact patterns while sourced but irrelevant answers stay unhealthy; nothing is sent or auto-applied.

Status
Loading review backlog…
Loading KB/instruction change tracking…
No-send review only; suggestions are not auto-applied.
Loading improvement backlog…

Improvement Candidates

Equivalent tenant-local findings are clustered by intent, answer route, and root cause. Every candidate remains a draft requiring human approval.

Loading Improvement Candidates…

AI Chat Logs

Runtime answer log with model/runtime, response ID, visitor question, answer preview, relevance, root cause, and handoff flags.

Status
Loading AI logs…
Loading AI chat logs…