I'd like to send an initial "working on it" message back to the user, then call our LLM to generate text. Is this possible? It looks liek you have to return the entire response

Yes, Custom LLM supports streaming, and you can send an initial "working on it" message before streaming the full response. Here's a code snippet to illustrate this pattern:

typescript
res.setHeader("Content-Type", "text/event-stream");
res.write(`data: ${JSON.stringify({ delta: "Working on it..." })}\n\n`);
// Call your LLM and stream its output
for await (const chunk of openAIResponse) {
  res.write(`data: ${JSON.stringify(chunk)}\n\n`);
}
res.write("data: [DONE]\n\n");
res.end();

This lets you immediately inform the user, then stream the LLM's response as it becomes available. For more details, see the [Vapi Custom LLM tool-calling integration guide](https://docs.vapi.ai/customization/tool-calling-integration). Source: - [Custom LLM Tool Calling Integration](https://docs.vapi.ai/customization/tool-calling-integration)