Deepgram acquires Daily and commits to keeping Pipecat open source

Deepgram announced the acquisition on October 29, 2026, at its inaugural Deepgram Speak event in San Francisco.[1] Daily’s team will join Deepgram, and Daily co-founder Kwindla Hultman Kramer will become Deepgram’s Head of Open Source.[3] Daily says Deepgram will continue to sell, support and develop all of its current products and services, and that nothing will change for Daily customers or the Pipecat ecosystem.[4] The companies did not disclose the acquisition amount.[2] Daily and Deepgram were in the same Y Combinator batch in 2016, and Deepgram has powered transcription inside Daily’s audio-video APIs since 2019, according to Daily.[4]

Deepgram says Pipecat will remain open source and free under its existing BSD V2 license.[3] It describes Pipecat as the world’s most popular open-source voice agent framework.[3] Daily says Pipecat has more than 200 integrations and almost 400 contributors.[4] Deepgram and organizations that contribute to and use Pipecat are establishing an independent Pipecat Foundation.[3] Daily says the foundation will be managed by a steering committee representing the realtime AI ecosystem, including companies that compete with Deepgram.[4] Deepgram says developers can choose and combine models, tools, network transport and APIs from different providers.[3]

Deepgram says Pipecat and Daily’s network infrastructure will underpin its new Voice Agent Platform alongside Deepgram’s speech models.[3] It describes the combined offering as the first complete, open, production-grade platform for conversational voice AI.[3] Daily’s global real-time network has 75 points of presence, and Deepgram says it connects end users to voice agents with ultra-low latency and high reliability.[3] TechCrunch describes Pipecat as managing voice and video streams and pipeline components such as speech-to-text, large language models and text-to-speech while reducing latency.[2] Deepgram also said its Chief Product Officer Ed Anuff would preview the forthcoming platform at Speak.[3]

At Speak, Stephenson introduced the Real-Time Context Standard (RCS), an open protocol co-developed with the Pipecat team.[1] Deepgram says audio context such as tone, speaker identity and background events can be lost as words pass between models.[3] RCS defines metadata for that context, instructions for how a language model should use it, and guidance for how a response should sound.[3] Its adapters translate context signals and delivery instructions for different models, allowing developers to use the same approach across providers.[3] Deepgram invited developers, model providers and voice-app teams to test RCS, provide feedback and contribute to its specification.[3]