Deepgram Acquires Daily And Pipecat To Build Open Production-Ready Voice AI Platform

Deepgram has agreed to acquire Daily, the company behind the open-source Pipecat voice agent framework, as it builds a broader production platform for conversational voice AI.

The acquisition brings Daily’s real-time communications infrastructure and Pipecat’s voice agent orchestration framework together with Deepgram’s speech models and forthcoming Voice Agent Platform.

Daily operates a global real-time network with 75 points of presence designed to support the latency and reliability requirements of live voice and video AI.

Pipecat, meanwhile, has developed into a widely used open-source framework for production voice AI, with integrations spanning major model providers, telephony systems, and deployment environments.

Following the acquisition, Daily’s team will join Deepgram.

Pipecat will remain free and fully open source under its existing BSD V2 license. Deepgram and other organizations involved in the Pipecat ecosystem are also establishing an independent Pipecat Foundation.

Deepgram said it will continue investing in Pipecat’s development and community while maintaining its vendor-neutral architecture. Developers will remain able to combine models, APIs, networking infrastructure, and other tools from different providers.

Kwindla Hultman Kramer, co-founder of Daily, will join Deepgram as Head of Open Source.

The acquisition also supports Deepgram’s broader effort to create a complete infrastructure stack for conversational voice agents, from speech recognition and speech generation to orchestration, networking, and deployment.

Deepgram is simultaneously introducing the Real-Time Context Standard, an open protocol developed with the Pipecat team.

The standard is intended to preserve information that can be lost as a voice interaction passes between different models and systems, including tone, speaker identity, unclear speech, and background context.

RCS defines metadata describing audio context, instructions for how a language model should interpret those signals, and guidance for how the resulting response should sound. Adapters can then translate that information for models from different providers.

Deepgram said the approach could allow voice agents to react more naturally, such as asking users to repeat unclear speech or responding quietly when someone whispers.

The company plans to integrate Pipecat and Daily into its new Voice Agent Platform while preserving developers’ ability to choose their preferred models.

Deepgram currently provides production voice AI infrastructure to more than 5,000 organizations globally. More than 650,000 developers have used its technology to process more than 200,000 years of audio.

KEY QUOTES:

“Keeping Pipecat open was fundamental to our decision to join Deepgram.”

“Deepgram has always been the voice AI company that takes production seriously, and their commitment to keeping Pipecat open is rooted in a belief, which I share, that the future of voice depends on developers being free to combine the best ideas from across the industry and build on each other’s work.”

Kwindla Hultman Kramer, Co-Founder Of Daily And Incoming Head Of Open Source At Deepgram

“Kwin and the Daily team deeply understand what it takes to keep a live conversation working reliably, and Pipecat gives developers the freedom to build with the models they choose.”

“Acquiring Daily allows us to bring that expertise together to tackle the whole voice system and give developers a faster path from prototype to production.”

“Too many teams treat voice AI as a feature to add late in the roadmap. In reality, it’s a distinct infrastructure discipline, with its own demands around latency, accuracy, and scale.”

Scott Stephenson, CEO And Co-Founder Of Deepgram