Pipecat is an open-source Python framework for building real-time voice and multimodal conversational agents, maintained by Daily and the community. It orchestrates audio and video, AI services and transports in a pipeline, so you can focus on what makes your agent unique instead of wiring together speech recognition, language models and speech synthesis yourself. You can build a single voice assistant or a full multi-agent system where specialists hand off, fan out in parallel and coordinate over a shared bus. Pipecat has more than 16,000 stars on GitHub and is actively developed in 2026.
Pipecat’s pipeline model streams audio and video frames through interchangeable processors: capture, voice activity detection, speech-to-text, an LLM, text-to-speech and a transport such as WebRTC. Each stage is swappable, so you can combine services from different providers, run locally or in the cloud, and add features like interruption handling and conversation context without restructuring your app. Beyond single agents, Pipecat supports multi-agent topologies that run locally or distributed across processes and machines, and a quickstart CLI scaffolds a working agent in minutes. Typical builds include voice assistants, AI companions, coaches and meeting assistants.
Pipecat the framework is free and open source under the BSD-2-Clause license. Daily also offers Pipecat Cloud, a hosted platform for deploying and scaling Pipecat agents, on paid plans, so the overall offering is freemium. You also pay for any third-party AI services you plug in.
Pipecat is for developers building voice-first products, realtime AI companions and multimodal interfaces who want an open framework instead of a closed voice-agent platform, with an optional hosted cloud when it is time to scale.