Pipecat

In this catalog it is the Voice Agents option when you need to turn speech into structured agent turns and back. Keep Pipecat on the observe/propose side of the loop: useful signals and drafts, with human or policy authority still required before anything irreversible runs.

Pipecat composes streaming voice/multimodal pipelines: STT → LLM → TTS over pluggable transports.

1. Define a pipeline of frame processors.
2. Transport carries audio/video frames.
3. STT/LLM/TTS stages stream results.
4. Output frames return to the user transport.

Related: livekit-agents · vocode · whisper · topics/03-channels

• Use when: Pipecat is an open framework for voice and multimodal agent pipelines — composing STT, LLM, TTS, and transports with streaming semantics matches the bottleneck and you specifically need to turn speech into structured agent turns and back.
• Skip when: another pack already covers the need with less ops surface, or Pipecat would become a second place people retype truth.
• Prefer instead: compare LiveKit Agents, Vocode, Whisper before locking the cell.

• Ops: confirm license, hosting, and release cadence for Pipecat (pipecat-ai/pipecat) before relying on it in a live corridor.
• Integration: bad credentials, scopes, or schema drift produce empty retrievals or runaway tool/model spend without an obvious UI alarm.
• Scope: Voice Agents components support agents; they are not the company ledger or the approval authority.