Automation Runs

See every automation run

Automated work is difficult to manage when all you have is a success or failure status. A status does not explain what the agent tried, what it produced, or why it stopped.

Automation Runs show the sessions created by schedules and subscriptions. Each run identifies the automation that started it and keeps the conversation as the primary record of the work.

Open the complete session

Open a run to read the conversation in order. You can review the agent's responses, tool activity, generated files, and requests for human input in the context where they happened.

From the run, you can also open the automation configuration or inspect the exact delivery that started the session. You do not need to search through the automation's complete delivery history.

Provide human input

Some automated work needs a person to complete a form, connect a service, or approve an action. These requests remain attached to the session.

Open the run, review the surrounding conversation, and respond to the request. Runs waiting on supported interactions can continue after you provide the required input.

Debug the result

The run keeps the evidence you need to understand what happened. You can inspect the conversation, tool inputs and outputs, errors, files, automation configuration, and originating delivery from one connected history.

This makes it possible to distinguish a temporary missing approval from a problem in the automation's instructions or configuration.

Improve the next run

After reviewing a run, open the originating automation and adjust its instructions or configuration. The current run provides the evidence for the change, while the automation remains the source for future runs.

Agenta does not change the automation automatically from your feedback. You decide what to update so similar runs need less intervention next time.

Learn how to create an automation, schedule an automation, or trigger an automation from an app.