Pipecat
Python framework for real-time voice and multimodal AI agents with pluggable speech, LLM and transport services.
Compose speech, LLM and video services into low-latency pipelines, locally or on Pipecat Cloud.
Pipecat is an open-source Python framework for real-time voice and multimodal conversational agents, maintained by Daily. It suits developers building phone bots, voice assistants and multi-agent systems who want to swap providers freely.
What it is
You compose a pipeline of services - speech-to-text, an LLM, text-to-speech, plus transports like Daily WebRTC or WebSockets - and frames stream through it so the agent starts replying while you are still talking. It adds interruption handling, turn detection, Pipecat Flows for scripted conversations, and multi-agent handoff.
Who it's for
It suits Python developers building voice agents for support, intake, coaching or phone calls. It does not suit non-developers, and you pay each speech, LLM and telephony provider you plug in.
Key features
Composable pipelines
Build agent behaviour from modular processors that stream frames in real time.
Large service catalogue
STT, LLM, TTS, speech-to-speech, video avatars and memory services from many named providers.
Pipecat Flows
Structured conversations with state management, including YAML-defined flows such as a restaurant booking.
Multi-agent workflows
Agents hand off, fan out in parallel, or run as sidecars over a shared bus.
Client SDKs
JavaScript, React, React Native, Swift, Kotlin and C++ SDKs connect apps to a Pipecat bot.
Project CLI
pipecat init scaffolds a phone or web bot project, with Claude Code skills for scaffolding and deploying.
Up and running in 3 steps
Our honest verdict
What's great
- Provider-neutral: swap speech, language and vision services, usually in one line, per the site.
- Frame streaming means no stage waits on the one before, which keeps replies fast.
- Includes ready examples for phone agents, IVR navigation and Twilio telephony.
- BSD-2-Clause licence is permissive for commercial products.
Where it falls short
- You need Python 3.11 to 3.14 (3.12+ recommended) and API keys for every STT, LLM, TTS and transport service you choose.
- Only the core framework installs by default; each provider needs its own extras package.
- Real phone and WebRTC deployments need telephony or transport accounts and hosting, which is more than a pip install.
- The Pipecat site says the team is joining Deepgram while staying open source, so governance may change.
Pricing & license
Pipecat is free under the BSD-2-Clause license. The README mentions deploying to Pipecat Cloud, but no prices are shown on the pages read, so the only cost named is running it yourself plus your STT, LLM and TTS providers.
Open remote AI-training roles
768 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
What Python version is required?
Python 3.11 through 3.14; 3.12 or later is recommended.
Can I deploy it myself?
Yes, run it locally or on your own servers, and the README also points to Pipecat Cloud for deployment.
Does it include a telephony provider?
It includes serializers and examples for providers such as Twilio, Exotel, Genesys and Plivo, but you bring your own account.
