Deepgram Is Buying Daily, Maker of the Pipecat Voice-Agent Framework, and Says Pipecat Will Stay Open
Speech-model company Deepgram said on Oct. 29 that it is acquiring Daily, the real-time infrastructure company behind the open-source Pipecat voice-agent framework. Pipecat stays under its BSD licence and moves to an independent Pipecat Foundation, Deepgram says, and Daily's team joins Deepgram. Terms were not disclosed. Deepgram also introduced an open Real-Time Context Standard and previewed a Voice Agent Platform built on Pipecat.
Deepgram is buying the open-source layer that many voice-agent builders use to wire together other companies' models, and its promise to keep that layer neutral is now the thing to watch. Deepgram said on October 29 that it is acquiring Daily, the company behind Pipecat, which Deepgram calls the world's most popular open-source voice-agent framework. Terms were not disclosed, TechCrunch reports.
The deal#
Pipecat will stay open source and free under its existing BSD V2 licence, Deepgram says, and the companies are setting up an independent Pipecat Foundation with organizations that contribute to and use the framework. Daily's team joins Deepgram, and Daily co-founder Kwindla Hultman Kramer becomes Deepgram's head of open source. "Keeping Pipecat open was fundamental to our decision to join Deepgram," he said in the announcement.
Daily also runs real-time network infrastructure with 75 points of presence, which Deepgram says will underpin its new Voice Agent Platform alongside its own speech models. Chief product officer Ed Anuff was set to preview the platform at Deepgram's Speak event on October 29.
The standard#
Deepgram also introduced the Real-Time Context Standard, an open protocol co-developed with the Pipecat team. It defines metadata for audio context, such as a whisper or an unclear word, instructions for how a language model should use those signals, and guidance for how a response should sound, with adapters for models from different providers, Deepgram says.
TechCrunch reports that Daily had raised more than $62 million, citing PitchBook, and that Deepgram became a unicorn with a $130 million Series C earlier this year. Both companies went through Y Combinator in 2016, according to TechCrunch.
What neutrality would have to cover#
Pipecat's code shows what Deepgram's promise involves. The framework's services directory, checked on November 2, holds 63 integration folders. They include Deepgram's direct speech-recognition competitors, such as AssemblyAI, Speechmatics, Gladia and Soniox, and voice-synthesis companies such as ElevenLabs and Cartesia. After the deal closes, Deepgram's staff will maintain the code that sends developers' audio to its rivals. Whether those integrations keep pace with Deepgram's own is the test of neutrality, and the open repository will make any difference visible.
The project is widely used by the measure GitHub provides. On November 2 Pipecat had about 17,000 stars and 3,000 forks under a BSD-2-Clause licence, against about 15,000 stars for LiveKit's Agents framework, which is licensed under Apache 2.0. Stars measure developer interest, not production use. On that measure, Deepgram's "most popular" claim holds against its closest open-source rival, but not by much.
Pipecat is one of several open routes to building voice agents; Mistral, for example, built its new voice-agent SDK on LiveKit. Vendor neutrality matters to developers who combine speech, language and voice models from different companies, and those developers now depend on a speech-model vendor keeping its word.
Limits#
- Price and terms were not disclosed, and the deal's closing conditions were not described.
- The Pipecat Foundation's governance, membership and funding have not been published.
- "Most popular" is Deepgram's description of Pipecat. GitHub stars and forks are the only comparison available here, and they do not measure production usage.
- The neutrality section is analysis; the integration count is a snapshot of the repository on November 2.