Deepgram Buys Daily, Maker of Pipecat, and Sets Up an Independent Pipecat Foundation
The voice AI company is keeping the open-source voice agent framework under its BSD license and introduced an open protocol for carrying audio context between models.
Published
Last updated / Fact-checked on
Deepgram said on October 29, 2026, that it is acquiring Daily, the company behind Pipecat, an open-source framework for building voice AI agents. Pipecat will stay open source under its existing BSD license, and the companies are setting up an independent Pipecat Foundation. Deepgram did not disclose the price.
Image: Illustrative photo of a condenser microphone. Kelly Sikkema (Unsplash) / CC0 1.0, via Wikimedia Commons
The deal
Deepgram, which builds speech recognition and voice AI infrastructure, announced the acquisition at Deepgram Speak, its event in San Francisco. Daily brings two assets: Pipecat, which Deepgram calls the world's most popular open-source voice agent framework, and a real-time network that Daily says spans 75 points of presence.
Following the acquisition, Daily's team will join Deepgram. Co-founder Kwindla Hultman Kramer becomes Deepgram's head of open source.
"Keeping Pipecat open was fundamental to our decision to join Deepgram," Hultman Kramer said.
Deepgram CEO Scott Stephenson said combining Daily's real-time expertise and Pipecat's model-agnostic design with Deepgram's models would give developers "a faster path from prototype to production."
What happens to Pipecat
According to the announcement:
- Pipecat stays fully open source and free under its BSD V2 license.
- It remains vendor-neutral, so developers can mix models, tools, network transport and APIs from different providers.
- Deepgram will invest in its development and community.
- An independent Pipecat Foundation is being formed with many of the organizations that contribute to and use the framework.
The release does not name the foundation's members, leadership or governance rules. Those details will decide how much independence the project actually keeps.
A new protocol: Real-Time Context Standard
Stephenson also introduced the Real-Time Context Standard (RCS), an open protocol co-developed with the Pipecat team. It targets a common gap in voice systems: words get passed from one model to the next, but audio cues such as tone, who is speaking and background noise get lost.
RCS defines three types of metadata:
| Metadata | Example |
|---|---|
| Context from the audio | A whisper or an unclear word |
| Instructions for the language model | How to use those signals |
| Delivery guidance | How the response should sound |
That lets an agent ask for clarification when it did not hear clearly, or answer quietly when someone whispers. Adapters translate the signals for each model, so the same approach works across providers. Deepgram is inviting developers and model providers to test RCS and contribute to the specification.
The bigger platform
The acquisition feeds into a forthcoming Deepgram Voice Agent Platform that combines Deepgram's speech models with Pipecat and Daily's network. Chief Product Officer Ed Anuff was scheduled to preview it at Speak, along with two new models for live conversation, expanded language support and a Global API aimed at improving reliability and latency. No launch date was given.
Deepgram says more than 5,000 organizations use its voice AI in production and more than 650,000 developers have used its tools.
Why it matters
Building a voice agent usually means stitching together speech recognition, a language model, speech synthesis, orchestration and a real-time network. With this deal, one company now owns the speech models, the most widely used open orchestration layer, by Deepgram's account, and the network that carries the audio.
That can simplify life for developers, but it raises a fair question about neutrality, because Pipecat integrates with Deepgram's competitors. The foundation and the BSD license are Deepgram's answer.
If you build on Pipecat, watch for the foundation's charter and membership. Check that integrations with other speech providers keep getting maintained. Try RCS in a test project before depending on it.
More coverage of voice AI is in our Artificial Intelligence hub.
How we reported this
We checked the deal, the licensing, the foundation, RCS and the company figures against Deepgram's announcement. Popularity and usage claims are attributed to Deepgram because we found no independent data. The release does not disclose deal terms or foundation governance.
Sources
DailyMech used AI tools to help research and draft this article; it was fact-checked by our editors.