I can help with this. The issue is that Vapi tool calls are synchronous by default, so your assistant expects a response immediately—returning a 200 ACK and replying later won’t work unless you design around that async gap. Typically, you’d handle this by either sending a quick fallback response (like “let me check that for you”) while your LLM processes in the background, then pushing the result into a follow-up interaction, or by using a callback/webhook pattern if your setup supports re-engaging the assistant. Quick question are you trying to have the assistant stay on the same call and respond once the LLM finishes, or are you okay with sending the result in a follow-up message or call?