Here's a nice tutorial on using MCP to extend the capabilities of a voice agent

July 31, 2025

Here's a nice tutorial on using MCP to extend the capabilities of a voice agent.

These days, lots of useful tools are exposed as MCP servers. It's great to be able to use these tools from a voice agent.

As usual, the unique thing about writing code for conversational voice is that response latency really, really matters.

My very general advice when people ask me about voice + MCP is to definitely use MCP for prototyping. After you've prototyped a feature that relies on MCP, it's often worth thinking about whether you should "unpack" the MCP functionality into manually written tool calls.

Writing tool calls manually is more work and more code to maintain, but gives you more leverage over latency and can help make your workflows more deterministic.

@pipecat_ai's MCP support is integrated with tool use (function calling) in general. So it's easy to use both MCP servers and hand-written tool calls. You can mix and match MCP tools, tools you implement in your code, and model-provided tools like Gemini's search grounding.

cerebriumai@cerebriumai

Ever wished your voice assistant could actually do something useful—like send invoices or manage subscriptions?

We just published a tutorial on integrating @PayPal's Model Context Protocol (MCP) into a real-time voice agent.

#mcp #voiceai #genai #llm