Pipecat Voice Agents Now Run on Telnyx Edge

Pipecat, one of the most popular open source voice agent frameworks, now runs as a function on Telnyx Edge. Deploy with telnyx-edge ship, connect a phone number, and call your agent: no servers, containers, or media infrastructure to run.

What's new

• Pipecat on Telnyx Edge: Deploy any Pipecat voice agent to an Edge function. Telnyx builds it, hosts it, and connects your callers to it.
• No framework changes: It's the Pipecat you already know with no new SDK. Your agent lives in function/bot.py.
• One platform, one API key: The phone number, call audio, Flux speech-to-text, LLM inference, and text-to-speech are all Telnyx.
• Available everywhere Edge runs: Pipecat agents work in every Telnyx Edge location, with Python 3.11 and WebSockets as the default runtime.

Why it matters

A Pipecat agent usually means stitching vendors together: one for the phone connection, one for transcription, one for the model, one for the voice, plus a place to host the bot itself. Deploying to Telnyx Edge removes the stitching. Every stage of the conversational loop runs on one platform with one API key, and the agent runs where calls land, so audio stays on the same infrastructure that terminates it. Bring the framework you already know; Telnyx handles the rest.

Example use cases

• Voice AI developers who want a production phone number in front of their Pipecat agent without standing up media infrastructure.
• Teams going from repo to callable agent with a single CLI command.
• Builders swapping providers: replace any Pipecat service and add its key with telnyx-edge secrets add.

Getting started

1. Create a Telnyx account and copy your API key.
2. Install the telnyx-edge CLI.
3. Clone the repo and create your function: telnyx-edge new-func -l python -n my-pipecat-agent --from-dir pipecat-edge.
4. Add your key and ship: telnyx-edge secrets add TELNYX_API_KEY "$TELNYX_API_KEY", then telnyx-edge ship.
5. Create a TeXML application that routes calls to your function's URL, connect a phone number, and call it.

pipeline = Pipeline([
    transport.input(),   # Telnyx call audio
    stt,                 # TelnyxSTTService (Flux)
    user_aggregator,
    llm,                 # OpenAILLMService on Telnyx inference
    tts,                 # TelnyxTTSService
    transport.output(),
    assistant_aggregator,
])

Learn more in the Pipecat on Telnyx Edge repo and the Telnyx Edge repo.