The short version
The most popular open-source layer for building voice agents now has a corporate owner, but its code is moving to a neutral foundation. Deepgram is buying Daily, the company behind Pipecat, and will run Pipecat through an independent Pipecat Foundation under the same BSD license. Teams that build AI agents with a voice interface keep their framework, yet should review how tightly their stack is tied to any one speech vendor.
What did Deepgram announce?
Deepgram, which sells speech-to-text and text-to-speech models through an API, said it will acquire Daily, a real-time video and audio infrastructure company that created Pipecat. TechCrunch reports that Daily had raised more than $62 million and that Deepgram raised a $130 million Series C at a $1.3 billion valuation in January 2026. Neither side disclosed the price.
Daily co-founder Kwindla Hultman Kramer joins Deepgram as Head of Open Source. Deepgram says Pipecat and Daily’s network, which it describes as 75 points of presence built for low-latency live AI, will sit underneath its new Voice Agent Platform alongside its own speech models.
Why does Pipecat matter to developers?
A voice agent is a pipeline: audio comes in, speech-to-text turns it into words, an LLM decides what to say, text-to-speech answers, and all of it has to happen fast enough that the conversation feels natural. Pipecat is the Python framework that wires those steps together, handles streaming and interruptions, and lets teams swap providers. Daily calls it the most widely used framework for voice agents and multimodal AI, with more than 200 integrations across model vendors, telephony and hosting.
That provider-agnostic design is why the ownership question matters. If the orchestration layer quietly favours one vendor’s models, switching costs rise for everyone built on it.
How will the Pipecat Foundation work?
Daily says the foundation will have a steering committee drawn from the wider real-time AI ecosystem, including companies that compete with Deepgram. Founding members and voting rules have not been published yet. TechCrunch quotes Kramer saying Deepgram CEO Scott Stephenson was committed to Pipecat being “open source and vendor-neutral forever,” and Stephenson saying the goal is to let everybody build voice agents.
The companies also introduced the Real-Time Context Standard, an open protocol for sending audio context, LLM instructions and response guidance to voice models. It is new and unproven, so treat it as something to watch rather than adopt immediately.
What it means for US & EU software teams
Voice is becoming a standard channel for customer service, scheduling, intake and field operations, and the market is consolidating around a few full-stack platforms. TechCrunch lists LiveKit, Vapi and ElevenLabs among the competitors. For buyers, the risk is less about Pipecat disappearing and more about drifting into a single vendor for models, transport and orchestration without deciding to.
Regulated products need extra care. Voice data is personal data under GDPR and can be protected health information under HIPAA, so where audio is routed, which regions process it and how long transcripts are kept should be explicit in your architecture and contracts, not inherited from a platform default.
What should teams building voice agents do now?
- Map your dependencies. List which vendor handles transport, speech-to-text, the LLM and text-to-speech, and which parts are Pipecat versus vendor SDKs.
- Keep providers swappable. Hide each model behind an interface and test at least one alternative for speech-to-text and text-to-speech.
- Pin versions. Lock Pipecat and its service plugins, and read release notes as governance moves to the foundation.
- Measure latency end to end. Track time to first audio and interruption handling per provider, so any switch is a data decision.
- Check data flows. Confirm regions, retention and subprocessors for audio and transcripts against GDPR or HIPAA requirements.
- Watch the foundation. Membership and decision rules will show how neutral Pipecat really stays.
Frequently asked questions
What did Deepgram announce?
On October 29, 2026, at its Speak event in San Francisco, speech AI company Deepgram said it is acquiring Daily, the maker of the open-source Pipecat voice agent framework and a real-time WebRTC network. Financial terms were not disclosed.
Will Pipecat stay open source?
Yes, according to both companies. Pipecat remains free under its existing BSD 2-clause license, and Deepgram is setting up an independent Pipecat Foundation with a steering committee that Daily says will include other companies in the real-time AI ecosystem, Deepgram competitors among them.
What happens to existing Daily customers?
Daily says Deepgram will sell, support and develop all of its current products, including WebRTC, telephony, the video and audio SDKs, live streaming, recording and Daily Prebuilt, with no changes for existing customers announced.
Should we change our voice agent architecture because of this deal?
Not urgently, but it is a good moment to check vendor coupling. Keep speech-to-text, the LLM and text-to-speech behind swappable interfaces, pin Pipecat versions, and track the foundation's governance before betting on any single vendor's bundled platform.
Sources
Deepgram — Deepgram is acquiring Daily, makers of Pipecat (primary source, October 29, 2026)
Daily — Deepgram is acquiring Daily and establishing an independent Pipecat Foundation (primary source, October 29, 2026)
TechCrunch — Deepgram acquires Pipecat-maker Daily to bolster its voice AI infrastructure (October 29, 2026)
The Next Web — Deepgram buys Pipecat maker Daily and keeps the voice AI framework open (October 2026)