🛠️ How-to guide

<aside> 📜

WHAT IT DOES

The Log tab shows every real conversation your agent has had. Replay them turn by turn to see exactly what happened — what the resident asked, what tools ran, and where things went well or wrong. Part of Inside Your Agent.

</aside>

Before you start

Conversations appear here once the agent is published and residents have interacted with it.

The conversations list

The left pane lists conversations with a running count. Each row shows a preview of the first message, how many messages and tool calls it involved, its duration, and how long ago it happened. Pin the ones you want to keep handy to the top, and page through the rest (20 at a time). Refresh to pull the latest.

The Log tab — the conversation list with message and tool-call counts, durations, and pinned favourites; select one to replay it turn by turn.

Replaying a conversation

Select a conversation to replay it turn by turn on the right — each resident message, the tool calls the agent made, its replies, and a per-turn pass/fail and duration. Handoffs to other agents or to a person show as dividers in the timeline. From here you can Analyze the conversation or Export it — Download report (.html) for a readable transcript, or Download data (.json) for the raw payload.

What to look for

• Failed turns — where the agent couldn't answer; open the turn to see why.
• Tool calls that didn't return what you expected — a sign to adjust an Advanced Configuration or knowledge connection.
• Escalations — conversations that went to a person, so you can spot gaps to close.

Use what you learn to refine the agent, then re-test before publishing.

FAQ

• I don't see any conversations. The agent needs to be published and used by real people first.
• Can I search or filter? Today you can pin favorites and page through the list, and replay any conversation in full; there's no free-text search in this view yet.
• Can I export a conversation? Yes — a readable .html report or the raw .json data.

See also

• Inside Your Agent