Know what’s working — and what to fix next

Overview

Metrics that drive better support

• Deflection at a glance

  See what share of conversations Gup resolves without a human — the metric that proves the agent is earning its keep.

• Knowledge diagnostics

  A queue of low-confidence or ungrounded AI replies — open the transcript and turn gaps into draft learnings.

• Action analytics

  Success, verify failures, errors, and latency for live Actions so you know Shopify order lookup (and future tools) are healthy.

• Close the loop with Learnings

  Pair diagnostics with Learnings and knowledge updates so every week of traffic makes the agent sharper.