Currently in our Vapi account, the First Message Mode is set to “Assistant waits for user” as per our client’s requirement. However, we are facing an issue where, in some outbound calls, the user remains completely silent after the call connects. In this scenario, Vapi continues waiting for user input without taking any further action. Our client has requested the following behavior: If there are 10 seconds of silence after the call connects, Athena should say: “Hello? I just want to make sure you can hear me.” If there is still silence for another 5 seconds, Athena should say: “Hi, this is Athena calling on behalf of [Company Name]. Can you hear me?” If total silence reaches 20 seconds, Athena should say: “It looks like we may be having a connection issue. I’ll disconnect for now. Thank you and have a great day!” After this message, the call should automatically disconnect. Could you please confirm whether it is possible to configure silence detection with timed prompts like this in Vapi? If so, kindly guide us on how to implement this behavior. Thank you.

Yes, this behavior is possible, but it isn’t handled automatically by First Message Mode alone. When the assistant is set to “wait for user,” Vapi will keep listening unless you add logic to handle silence. The usual approach is to use silence detection with timed actions, either by configuring silence timeouts in the assistant or by handling it through server events/webhooks that track elapsed silence after call connect. You can then trigger the assistant to speak predefined prompts at specific time intervals and end the call if silence continues. This requires a bit of custom logic, but it’s a supported and recommended pattern for exactly this use case.

Hi Azhar, Yes, this is possible. Please refer to the documentation below for full details: https://docs.vapi.ai/assistants/idle-messages Vapi supports the

customer.speech.timeout

event. This event is triggered when: • No user speech is detected • For a specified duration You can also chain multiple timeouts to create progressive follow-ups or escalation logic depending on how long the caller remains silent.