Skip to content
Conversation

Run a ticket. Watch every span land.

Loading active prompt...

Pick a scenario. The agent uses three tools (customer context, policy search, retrieve-similar-resolutions via Phoenix MCP) and one optional write (escalation). Every span is captured by Phoenix and rendered in the live trace pane on the right.

Each scenario maps to a row in the seeded SQLite. Live mode hits Gemini and Phoenix; set LIGHTWEIGHT_DEMO=true for fixture-only.

No ticket selected.

Pick a scenario above to load a customer message and run the agent against it.