Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer Based on the troubleshooting guide for Custom LLM integration, here are steps to diagnose your issue with the test audio button: [(1) ]() Step 1: Test Connection with Postman First, verify your WebSocket connection is working properly. Open Postman, create a new WebSocket request, enter your WebSocket URL, click "Connect", then navigate to the "Message" tab, enter "Hello", and click "Send". You should receive the message back in your server indicating a successful connection. Step 2: Verify Endpoint Protocol Ensure your URL protocol matches your connection type: - HTTPS Endpoints: Use

wss://

prefix for secure WebSocket connections - HTTP Endpoints: Use

ws://

prefix for standard WebSocket connections Step 3: Check for Trailing Slash Some server setups might not allow a trailing slash in the URL. Check your URL format carefully: - Correct Format: ` - Incorrect Format (sometimes) : ` Step 4: Check for Retell Server Response Add logging in your websocket code to see if the websocket is connected, and add logging in your websocket message code to check if you receive the message from Retell server. If the issue persists after these checks, I'd recommend reaching out for additional support through the community channels.

​ To learn more, see the sources I considered: [[1] Troubleshooting Guide](https://docs.retellai.com/integrate-llm/troubleshooting) ​

@AI Evy the test audio button doesnt work

### Answer Since you've confirmed that your WebSocket connection works and you've successfully connected Retell to n8n, but the test audio button isn't clickable, here are additional troubleshooting steps specific to Custom LLM integration: Verify Your Server Sends the First Message Your server needs to send the first message when the WebSocket connection is established [(1) ]() . According to the protocol, your server must send an optional config event and a response event that serves as the begin message for the agent to speak [(2) ]() . If you want the agent to wait for the user to start the conversation, set content to empty string [(1) ]() [(2) ]() . Check Server Compatibility Verify that your server setup supports WebSocket connections [(3) ]() . For example, Vercel edge functions and serverless functions cannot act as a websocket server . If you're experiencing automatic disconnections, check if your server vendor has timeout limitations . Add Server-Side Logging Add logging in your websocket code to see if the websocket is connected, and add logging in your websocket message code to check if you receive messages from the Retell server . Protocol Requirements The protocol requires [(4) ]() : - Your server to send the first message: send empty response to let user speak first - Retell will send live transcripts to your server, and expect responses when needed - You will stream what you want your agent to say to Retell's server, and it will speak it out If these steps don't resolve the issue, I recommend f

iling a bug report or reaching out to Retell support directly with your specific setup details.

​ To learn more, see the sources I considered: [[1] Setup WebSocket Server](https://docs.retellai.com/integrate-llm/setup-websocket-server) [[2] LLM WebSocket](https://docs.retellai.com/api-references/llm-websocket) [[3] Troubleshooting Guide](https://docs.retellai.com/integrate-llm/troubleshooting) [[4] Custom LLM Overview](https://docs.retellai.com/integrate-llm/overview) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

This is the root cause! Retell's Custom LLM mode expects a WebSocket connection but n8n webhooks are HTTP POST — they are completely different protocols. This is why Test Audio doesn't work — Retell is trying to open a WebSocket to your n8n URL but n8n only speaks HTTP. can i Switch Retell agent type to "Custom LLM - HTTP" ? @AI Evy

### Answer You've correctly identified the issue. Retell's Custom LLM integration requires a WebSocket connection, not HTTP POST [(1) ]() [(2) ]() . Unfortunately, there is no "Custom LLM - HTTP" mode available in Retell AI based on the available documentation. The Custom LLM integration specifically requires : - A WebSocket server that handles text exchanges with Retell server - Your server to send the first message upon WebSocket connection - Real-time bidirectional communication via WebSocket protocol Since n8n webhooks only support HTTP POST and cannot act as a WebSocket server, you have two options: Option 1: Build a WebSocket Server Set up a separate WebSocket server (using Node.js or Python) that acts as a bridge between Retell and n8n . This server would: - Handle WebSocket connections from Retell - Make HTTP POST requests to your n8n webhook when needed - Stream responses back to Retell via WebSocket Option 2: Use Retell's Agent Frameworks Instead of Custom LLM, use Retell's built-in agent frameworks (single prompt or conversation flow) which provide more capabilities and built-in tool sets [(1) ]() . These frameworks can still integrate with external systems via webhooks and custom functions. The Custom LLM integration is specifically designed for cases where you have specific compliance or use case requirements that necessitate hosting your own LLM .

​ To learn more, see the sources I considered: [[1] Custom LLM Overview](https://docs.retellai.com/integrate-llm/overview) [[2] Setup WebSocket Server](https://docs.retellai.com/integrate-llm/setup-websocket-server) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.