Running the Agent
The agent starts on its own when a ticket arrives from your ticketing system. It reads the ticket, pulls evidence from your codebase, logs, and documentation, and leaves a ready-to-send response here. To run it again, click Deploy. Do that after new replies come in, or when you want to add context the first run didn’t have. View details shows the reasoning as it happens, and Stop cancels a run. Once a ticket has replies, you can choose what the agent reads: Ticket only for the original description, or With replies for the whole thread.Reading the Answer
The Response tab holds two views of the same run:- Investigation: the technical breakdown for your team, with code references and root cause
- Response: the message you can send to the customer
- Sources: the citations behind the answer, including code, articles, and logs
- Reasoning: the agent’s investigation, step by step
- Copy the response to your clipboard
- Re-run the agent, adding instructions or attachments to steer the next investigation
- Save to Knowledge Base: turn the findings into an article
- Feedback: thumbs up or down, which feeds back into future answers