----------------------------- **While I generate an answer, here are some sources that might be helpful**: * [Outbound Sales Example]() * [Create Assistant]() * [Inbound Support Example]() * [Create Phone Call]() -----------------------------

### Answer It sounds like your assistant might be still configured with a server URL or a function that leads to the call being ended prematurely. When a

serverUrl

is specified in an assistant's configuration, Vapi will send function call information to this URL. If there's a function configured to end the call after a certain action (like scheduling an appointment), and the server handling the function calls doesn't properly manage the flow or erroneously confirms the action, the call might end unexpectedly. You should check the assistant's configuration to see if there is a

serverUrl

set and review the server-side logic associated with the function calls being made. If the

serverUrl

is set at the assistant level, it will override any account-level URL. This configuration can be reviewed and modified either through the Vapi dashboard or through API calls to update the assistant's settings. If you need to update or remove a

serverUrl

, you can use the Update Assistant endpoint. Here's an example of how to update the assistant to remove a

serverUrl

, which might be causing the unexpected behavior:

json
{
  "serverUrl": ""
}

This JSON payload can be sent to the Vapi API to update the assistant's configuration, ensuring that no server URL is involved unless explicitly needed [(1)]() . To make these changes through the API, you might use a command like the following:

bash
curl --request PATCH \
  --url https://api.vapi.ai/assistant/{assistant_id} \
  --header 'Authorization: Bearer {your_token}' \
  --header 'Content-Type: application/json' \
  --data '{"serverUrl": ""}'

Make sure to replace

{assistant_id}

and

{your_token}

with your actual assistant ID and authorization token [(2)]() . This should help resolve the issue if it's related to unintended function calls due to server URL configurations. If the problem persists, it might be beneficial to inspect the detailed logs for the calls using the assistant ID or call IDs you provided, to better understand what's happening during those interactions.

@Sahil I scheduled a meeting through your link but never got the e-mail

trying to record a demo and the assistant performance sux

Showderr tried to help me yesterday, we built a GPT that would call the API but it still won't work. tried this morning and it says that the assistant 234e76d8-89b4-4a8e-9af8-7c419b1b5483 does not exist