I'm testing an assistant I've created within my saas product and I continue to see the Success Evaluation returned as null when the call is longer than ~minute. On the other hand, the summary returns correctly populated regardless of the call length. Here are some failing call ids - 476e65b9-17de-4371-8858-1be1a8831934 - 3cb6ea6a-caf8-474a-a15a-132e26dfb7d6 - 79f438f2-9d33-480f-9729-f0152e17bea7 Other assistants have no problem returning the success evaluation with loger calls. Here are some working call ids: - a499f7e7-631e-41e2-95c3-b20836d2c4e2 - 6ae6aa7b-346b-4922-bfe3-8c779aa9bc2f - 0806629d-981c-41ca-99f7-669d5ca59f29 I'am using vapi for web calls and am building my assistant configuration on the fly and passing to the vapi.start() method. https://docs.vapi.ai/quickstart/web#option-1-temporary-assistant The failing assistant does have a longer system prompt so maybe that's the cause, but that's just a guess?

@certified dirtiboi Apologies for the delayed response. If we look at the logs and your call transcript you will see below the timeout occured. You just need to increase the timeout for generating analysis plan success evaluation and then it will work as it expected.

json
"successEvaluationPlan": {
  "rubric": "Checklist",
  "enabled": true,
  "messages": [],
  "timeoutSeconds": 60
}

logs

🟡 20:06:33:612 Couldn't Complete Anthropic Bedrock Request. Status: undefined. Error: Request timed out.

okay thanks, I'll try this out