My Agent Didn’t Run

You clicked Run Agent, but nothing happened. No output was returned, or you received a vague error message. This usually means something in the Agent setup was incomplete, misconfigured, or unable to connect to its required inputs.

What to check

Before you retry, run through this quick checklist:

• Are all required fields filled?
  Even one missing dropdown or input box can block the run.
• Is the Site active and correctly linked?
  Your Agent must belong to a live Site with an assigned Brand Voice.
• Is the LLM model available?
  Some models may be disabled or temporarily restricted. Try switching to GPT-4.1 or Claude 3.5 as a fallback.
• Does the Datastream return results? (if applicable)
  If your Agent relies on a Datastream, check that it isn’t returning zero content.
• Did your connection time out?
  A long wait with no output could be a temporary timeout issue.

How to fix it

Try the following steps:

1. Refresh the page and reopen the Agent. This clears temporary session issues.
2. Recheck your input fields — even optional fields might have logic dependencies behind the scenes.
3. Use Prompt Studio to isolate the issue. Paste in the prompt manually and test your inputs one at a time.
4. Switch to a more reliable model (like GPT-4.1) if you suspect model failure.
5. Preview your Datastream (if used) and confirm it's pulling in live content.
6. Duplicate the Agent and try again — sometimes cloning resets hidden conflicts.

If the Agent still won’t run and you’ve checked all the above, contact support and include:

• A screenshot of your Agent setup
• Which model and Site you were using
• The time and name of the failed run (if available)

We’ll help resolve the issue directly and ensure your Agent gets back to work.