[Urgent] Vapi is abruptly ending without completin...
# support
a
@Vapi @Sahil I'm having some issues that used to not happen with my outbound calls. Did something change recently? The problem: 1. We are starting to see Vapi fail to complete the tasks we instruct it to complete, namely asking a set of questions and then get the availability. 2. The Assistant abruptly says the endCall phrase "Goodbye. Bye" without collecting all the information over the phone. 3. The user responses are all streamlined with no edge cases, and Vapi still isn't able to get to the end. Example call ids that are failing to complete: 1.
c4ccabdf-da57-4731-88eb-860c790c0839
2.
e26456c1-7eca-4a50-bccd-7d9ce4e4e98f
3.
c694db92-14ed-41e8-81f1-492b156daa6f
More details on the prompt we use and the full object we send are in this support thread.
The prompt:
Copy code
[Identity]
You are an assistant of Ajay Vasisht. Your goal is to call and check-in with a patient since their last visit which happened in the last week.

[Style]
- Be concise and to the point.
- Personalize the interaction by including the caller's name throughout the conversation when natural.
- Maintain a professional and empathetic tone throughout the call.
- When it is time to end the call, make sure to say 'Thank you. Goodbye.'.

[Response Guideline]
- Keep responses minimal and focused on the task at hand.
- When asking questions, don't list the question number. Just ask the question.

[Task]
1. Wait until the caller answers the phone and greets you. If the caller doesn't say anything in 5 seconds, say your first message.
2. After the caller greets you, introduce yourself as ${assistantName}, an administrative assistant for Ajay Vasisht.
3. Clarify that you are calling to check in on them since their last visit which happened in the last week.
4. Ask the questions in the order they are listed in the '[Questions]' section.
5. When you are confident you have received answers to every question, say thank you and that you will message them to confirm. Then, end the call politely.

[Questions]
- How are you feeling now?
- Do you have any new symptoms since your last appointment?
- Have you been following the prescribed treatment plan?
- Do you need any clarifications about the last appointment?
- What's your availability for a follow-up appointment? Can you give the earliest available date and time?
The full object
CreateAssistantDTO
we send to Vapi's API for an outbound call: https://gist.github.com/avasisht23/4d508189dbcb2a991d87f700b4c9c6d3
v
message has been deleted
@Ajay Vasisht I'm sorry to hear that you're experiencing issues with your outbound calls. It's important to ensure that everything is set up correctly and that there are no underlying problems causing these failures. Let's take a closer look at the issues you've described. 1. For the tasks failing to complete, we need to check if there have been any changes in the system or the way the tasks are being handled. It's possible that there might be a bug or a change in the API that's affecting the functionality. 2. If the Assistant is ending calls prematurely, this could be due to an error in the call flow logic or an issue with the Assistant's configuration. We need to review the call flow and the conditions that trigger the endCall phrase to ensure they are set up as intended. 3. Even with streamlined user responses, if VAPI isn't able to get to the end, it suggests there might be a problem with the call logic or the Assistant's understanding of the responses. To address these issues, I will need to review the details provided in the support thread you mentioned. Since you've provided example call IDs, I can look into the logs for those calls to identify any common patterns or errors that might be causing these problems. Could you please provide me with the prompt you use and the full object you send, or direct me to the support thread where these details are available? This will help me to diagnose the issue more effectively.
You can continue this conversation by mentioning me in the message.
Got it. Tagging @User for additional help.
a
Hey @Shubham Bajaj @Sahil any thoughts here?
s
Can you check which model took the endCall decision? Check the logs during the end you will get the model name. I am guessing gpt-3.5-turbo must have been used and it hallucinated the response
a
@Sahil well I set the model in my JSON to be gpt-4o. Why does Vapi need to fallback to gpt-3.5-turbo?
s
@Ajay Vasisht can you try again, from 08-09 the model hallcuinated and you observed absurd bheaviour.
a
works now on a simpler prompt, ty
2 Views