Team Execution Records

View a record of every team run, including its conversation, the assistants involved, each assistant's token usage and duration, and filter all runs from the same conversation.

Teams automatically hand off and delegate work between assistants without interrupting the user. Execution records let you review which assistants handled each conversation and how many resources they used. Use them to confirm that the team divided work as expected and to troubleshoot and optimize it.

1. View execution records

On the team page, switch to Execution records to see every run for this team. Each message a user sends to the team creates one record. Each record includes:

Field	Description
Record ID	Unique identifier for this run.
Conversation	The conversation this run belongs to, shown by its title. Click the title to open All conversations and view the full conversation.
Status	Running, Completed, Failed, or Timed out.
Iterations	Total number of exchanges between assistants during this run.
Tokens	Total tokens used in this run (input in / output out).
Duration	Total time from receiving the message to completing the reply.
Created at	When this run occurred. Click the column heading to switch between oldest-first and newest-first sorting; the default is newest first.

Execution records: each record shows its conversation title, so you can see at a glance which records belong to the same conversation

Can't open a conversation title? You need access to Conversation records for titles to appear as links. Without access, titles appear as plain text (hover over a title to see why).

2. Filter execution records by conversation

A conversation usually contains several messages, each with its own execution record. To focus on one conversation, show only its records.

1. Filter by this conversation

In the Actions column to the right of the record you want to inspect, click the funnel icon Filter by this conversation.

Click the funnel icon to the right of a record to filter by this conversation

2. Review the filtered results

A banner above the list says Showing execution records for conversation “…” only. The list then shows only records from that conversation. The filter remains in place when you change pages or sort the list. The funnel icons for these records turn gray; hovering over them shows Already viewing records for this conversation.

To return to the full list, click Clear or the close icon (×) on the right side of the banner.

After filtering, the banner identifies the selected conversation and the list shows only its records

Example: Find which assistant answered a customer complaint

A customer service supervisor receives feedback that the return instructions were unclear and wants to find out which assistant handled each reply in that conversation:

1. Find any record from the conversation under Execution records and click Filter by this conversation. The list now contains only records from that conversation.

2. Expand each record. The assistant names on the step cards show that the first message was received by Demo – Customer Service Routing Assistant and handed off to Demo – Returns and Exchanges Assistant. The latter answered the follow-up directly.

3. Click the title in the Conversation column to open the original conversation and compare the actual replies. Once you confirm the issue came from Demo – Returns and Exchanges Assistant, return to AI assistants to adjust its settings.

3. Expand step details

Click + to the left of a record (or click the row) to see which assistant nodes handled the run, in order. Each step appears on a card:

Item	Description
Step number and name	Shown as #number · name, such as #1 · Demo – Customer Service Routing Assistant, so you can identify the assistant responsible for the step.
Token usage	The right side of the card shows the step's input/output tokens (in / out). Some steps also show duration, helping you identify where resources were spent.

Step details: each card shows its number and assistant name, with token usage on the right

No records visible? Execution records appear only after a conversation triggers the team. If the list is empty, chat with the team through its conversation platform, then return to this page.

4. Use records to optimize the team

• Unexpected routing: If the step cards show that a request went to the wrong assistant, return to Canvas composition and clarify the connection's Trigger condition description to specify when that route should be used.

• Too many iterations or a long duration: Assistants may be handing the request back and forth too often. Check for unnecessary connections, or adjust the team's Maximum iterations and each node's settings.

• Costs concentrated in one assistant: Compare token usage across the step cards to find the assistant consuming the most tokens. Consider using a lighter model or narrowing its knowledge base and tool scope.