Run an AI agent session
Use the Agent Runner to start, monitor, and control an AI agent as it works through tasks in real time.
This guide is designed for project manager.
Steps
Step 1: Open the Agent Runner
Navigate to the Agent Runner screen from your workspace navigation to access the live agent session view.
Step 2: Review the agent context and objectives
Check the agent’s configured goal, any input parameters, and the current context so you understand what it will work on.
Step 3: Start the agent run
Click the control to start or resume the agent so it begins processing the task, then confirm that the session status changes to running.
Step 4: Monitor agent activity in real time
Watch the live log or message stream to see each action the agent takes, including status updates, decisions, and any generated outputs.
Step 5: Inspect intermediate results
Open individual log entries or result panels to review details of the agent’s intermediate outputs, such as summaries or data it has produced.
Step 6: Pause or stop the agent
Use the pause or stop controls if you need to temporarily halt or fully end the current run, for example if objectives change or an issue appears.
Step 7: Restart or rerun with updated context
Adjust the agent’s inputs or context if needed and then start a new run to let the agent continue with the refined requirements.
Step 8: Review final outcome and status
Once the run completes, check the final status, outputs, and any summary provided by the agent to confirm that objectives have been met.
Last updated: 2026-09-01