Overview
The Activity log records every run of every AI agent on your account: which agent ran, in which project, what it did step by step, and whether it finished. This is where you look when an agent did something you did not expect, or did nothing when you expected it to act.
Read more about building and running agents in Create & Manage Your AI Agents.
Read more about Privacy & Security for AI Features.
Access & Permissions
The Activity log is available in the Webapp only. You need an in-house user with the user permission AI Agents. Read more in Permissions.
How to Access the Activity Log
To see the runs of all agents:
- Click AI Agents
- Click Activity log
There are alternative ways to access the activity log of an agent that you are working on. These two ways just navigate to the Activity log and set the filter for that agent.
To see the runs of one agent from the agent list:
- Click Agents (if you are not there already)
- Open the More menu of the agent
- Click Activity log
To see the runs of one agent from the agent screen:
- Click Agents (if you are not there already)
- Click the agent (if you are not there already)
- Open the More menu
- Click Activity log
Read the Activity Log
You see the runs of every agent that runs in a project you have access to, including agents a colleague created.
At the top of the page, four tiles summarise the runs currently listed: Total runs, Completed runs, Failed runs and Agents that ran:
Runs are listed newest first and grouped by day, under TODAY, YESTERDAY and then by date. Each row names the agent, its outcome, the project it ran in, how long it took and when it ran:
Every run carries one of four outcomes:
| Outcome | What it means |
| Completed | The agent finished the run. It may have created something, or decided that nothing was needed — both are completed runs. |
| Failed | The run stopped before the agent finished its work. |
| Running | The run is still in progress. |
| Deleted | The agent that made this run has been deleted. The row shows the name it had before. |
A run that reaches a limit, or that the security policy stops, is also a Failed run. Open the run to see which step was stopped.
A run stays in the log after its agent is archived or deleted.
Agents that keep failing pause themselves. Read more in Create & Manage Your AI Agents > When an Agent Pauses Itself.
Test runs are never recorded here — the Activity log only holds runs that really happened. Read more in Create & Manage Your AI Agents > Test the Agent.
Filter the Activity Log
To filter the list:
- Click the Filter icon
- Select an Agent, a Status or a Date
These are the Date options:
- Today
- Last 7 days
- Last 30 days
These are the Status options:
- Completed
- Failed
- Running
Runs of a deleted agent carry a Deleted badge but cannot be filtered for.
Each filter you set is shown as a chip above the list, and the four tiles recount for the filtered runs only.
Read the Execution Log of a Run
Every run keeps a step-by-step record of what the agent read, what it decided and what it created.
Only the person who created the agent and the account owner can open a run's Execution log. Everyone else sees that the run happened — the agent, the outcome, the project and when it ran — but cannot open it. The Account user permission does not grant access.
To see what an agent did in a run:
- Click the run
Under Execution log each step is written in plain language and names the real tickets and documents involved — the ticket the agent read, what it looked for, how many results it found, and the ticket or comment it created at the end.
A run can end without the agent doing anything. The log then closes with a line saying that the agent reviewed the event and decided no action was needed. This is a normal result, not a failure.
If the agent tried to do something it is not allowed to do, that step appears in the log as blocked. This line is written in the language of the person who created the agent, not in yours, so you may see it in another language than the rest of the log.
Comments
0 comments
Please sign in to leave a comment.