wise-white•2d ago
Assistant hook customer.speech.timeout running mid tool call
We are using the assistant hook on "customer.speech.timeout" to ask if the user is still on the line. The issue is that if a tool call is taking time, the assistant hook is still running asking if the user is there.
1 Reply
wise-whiteOP•2d ago
This is in the documentation
@Duckie Are you able to help with this?
It is actually not the tool call. It is if there is a model request that is in flight that is taking longer than the timeout, the timeout is hit. There are times where the server latency is greater than the timeout to get back the entire message
This still might be a vapi issue because we are streaming back the tool response, but it is taking longer than our timeout and vapi isn't thinking that it is processing those messges. In the log all we see is the request and the response, not the Model output logs