@VR To address the issue where Vapi pings multiple times for a user response, causing the agent to speak responses for each ping, you can implement a mechanism to track the latest request and ignore the stale ones. Here's a strategy you could use: 1. Maintain a variable that stores the ID of the latest request. 2. When a new request comes in, update this variable with the new request ID. 3. Before processing a request to speak, check if the request ID matches the latest request ID stored. 4. If it does not match, it means the request is stale, and you should not proceed with the

Speak()

function. 5. If it matches, proceed with the

Speak()

function. Here's how you could modify your endpoint to include this logic:

typescript
let latestRequestId: string = "";

app.post("/chat/completions", async (req: Request, res: Response) => {
    const requestId = uuidv4();
    latestRequestId = requestId; // Update the latest request ID

    req.on("close", () => {
        res.end();
    });

    // Check if the current request ID is the latest before speaking
    if (requestId === latestRequestId) {
        Speak(res, requestId, "Hello, how are you doing today? I'm doing fantastic - it's such a beautiful day outside");
    }

    res.end();
});

This should ensure that only the response to the latest request is spoken by the agent. If you encounter any further issues or need more assistance, please let me know!

You can continue this conversation by mentioning me in the message.

Got it. Tagging @User for additional help.