The voice AI sector continues to accelerate, and leading companies are consolidating through acquisitions and partnerships. On Thursday, Deepgram announced it is acquiring Daily, the developer and maintainer of Pipecat — a widely used open-source framework for real-time voice and video pipelines.
The companies did not disclose the transaction amount but confirmed that Pipecat will remain open-source.
What Pipecat does and why it matters
Pipecat is a framework that manages audio and video streams and connects them with other pipeline components such as speech-to-text (STT), large language models, and text-to-speech (TTS), while aiming to reduce latency. Daily also offers hosted infrastructure for any kind of voice and video calls.
For Deepgram, acquiring Pipecat brings deeper expertise in orchestrating real-time audio and video systems and provides a codebase and design patterns that can speed integration with Deepgram’s own offerings.
Funding background and ties
According to PitchBook, Daily has raised more than $62 million to date from investors including Tiger Global, Highfield Capital, Nat Friedman, Lachy Groom and Kevin Weil. Deepgram earlier this year reached unicorn status after a $130 million Series C round. Both companies participated in the same Y Combinator batch in 2016.
Why Daily agreed to the deal
Daily co‑founder Kwindla Kramer said the company began experimenting with AI‑to‑human voice and text interactions around the time GPT‑4 launched, which led to the creation of Pipecat. While multiple acquisition offers were reportedly received, Kramer favored Deepgram because CEO Scott Stephenson committed to keeping Pipecat open-source and vendor-neutral permanently, and because the engineering cultures of the two organizations were compatible.
Kramer also emphasized that Pipecat was designed not merely as a voice toolkit but as a foundation for building a new class of AI-native software that assumes real-time operation.
Deepgram’s goals and technical direction
Scott Stephenson said Deepgram had long wanted greater involvement in open source, and saw an orchestration framework for agents as a promising area for collaboration with Daily. Deepgram intends to open the ecosystem so anyone can build voice agents, whether or not they use Deepgram models, and is prepared to improve its own models where necessary based on ecosystem usage.
The acquisition enables developers to assemble expressive end‑to‑end speech systems by plugging in different models and using a context bus that connects system components.
New open protocol: Real‑Time Context Standard (RCS)
Deepgram and the Pipecat team also co-developed and are launching the Real‑Time Context Standard (RCS), an open protocol for voice agents. RCS defines metadata such as contextual signals extracted from audio (for example, a whisper), instructions for how a language model can interpret those signals, and guidance on how responses should sound.
Market context
Voice AI has attracted investment on both the model and infrastructure sides. Model developers such as ElevenLabs, Gradium and Fish Audio have raised millions, while infrastructure companies like LiveKit and Vapi have completed multi‑million‑dollar funding rounds this year.
The Deepgram–Daily acquisition is another step in consolidating real‑time voice and video AI infrastructure, combining technical capabilities with a commitment to maintaining openness in the ecosystem.
