Deepgram Acquires Daily and Open-Source Voice AI Framework Pipecat
Deepgram has acquired real-time infrastructure provider Daily, pledging to keep the Pipecat framework vendor-neutral under an independent foundation.
In short
- Deepgram has acquired Daily, the creator of the popular open-source voice agent framework Pipecat, for an undisclosed sum.
- Pipecat will remain free and open source under its BSD V2 license, supported by a newly established independent Pipecat Foundation.
- The companies introduced the Real-Time Context Standard (RCS), an open protocol designed to preserve audio nuances like whispering and tone across model pipelines.
- Daily co-founder Kwindla Kramer joins Deepgram as Head of Open Source, bringing the 18 engineers who built the Pipecat framework and 75 points of presence into Deepgram's infrastructure stack.
On October 29, 2026, voice AI infrastructure company Deepgram announced that it is acquiring Daily, the creator and maintainer of the open-source voice agent framework Pipecat. The acquisition unites Deepgram's automated speech recognition and voice models with Daily's real-time transport network and orchestration tooling. While financial terms were not disclosed, both companies stated that Pipecat will remain vendor-neutral and open source under an independent foundation.
The transaction brings together two companies that shared the same Y Combinator cohort in 2016, according to reporting by TechCrunch. Prior to the deal, Daily had raised over $62 million in venture funding from firms including Tiger Global and Highfield Capital. Deepgram achieved unicorn status earlier in 2026 following a $130 million Series C funding round.
Deepgram Acquires Daily to Unify Voice Infrastructure
Daily built its business around real-time communication infrastructure, maintaining a global network of 75 points of presence engineered for low latency. In 2023, following the release of GPT-4, Daily began building tools for voice agents, which led to the creation of Pipecat. Daily open-sourced the entire Pipecat stack in 2024, including the orchestration framework, client-side SDKs, the RTVI messaging standard, SmallWebRTC transport, and the Pipecat Smart Turn detection model.
Deepgram's transcription models have powered transcription inside Daily's audio-video APIs since 2019, according to Daily's announcement. Under the acquisition, Daily's commercial infrastructure products—including WebRTC, telephony connections, and cross-platform libraries—will be integrated into Deepgram, while Deepgram will continue to sell, support, and develop Daily's existing commercial services.
Daily co-founder Kwindla Kramer is joining Deepgram as Head of Open Source. Kramer noted that Daily favored Deepgram because of CEO Scott Stephenson's pledge to maintain Pipecat as an open and vendor-neutral technology.
The Independent Pipecat Foundation
To ensure Pipecat does not become a proprietary silo, the companies are establishing an independent Pipecat Foundation, as detailed by Deepgram's blog. The framework will remain available to all developers under its existing BSD V2 license.
The foundation will be governed by a steering committee representing organizations across the real-time AI sector, including direct competitors to Deepgram. Pipecat currently counts more than 200 integrations and nearly 400 contributors across the software ecosystem, supporting pipelines that connect speech-to-text, large language models, and text-to-speech services.
Stephenson emphasized that the open architecture is designed to accommodate multiple providers. Developers building on Pipecat will retain the ability to choose, swap, and integrate models, network transport systems, and third-party APIs rather than being locked into Deepgram's proprietary catalog.
Launching the Real-Time Context Standard
Alongside the acquisition, Deepgram and Daily unveiled the Real-Time Context Standard (RCS) at the Deepgram Speak event in San Francisco. RCS is an open protocol designed to solve a persistent issue in modular voice agents: the loss of non-verbal audio context.
In typical voice pipelines, spoken audio is converted to plain text before passing to a language model, stripping away acoustic markers. RCS defines three categories of metadata: audio context (such as an unclear word or a whisper), instructions directing the language model on how to interpret those cues, and delivery guidelines indicating how the generated response should sound. Model adapters translate these signals across various third-party models, allowing agents to respond softly to a whispered query or seek clarification when background noise interferes.
At the same event, Deepgram Chief Product Officer Ed Anuff previewed the Deepgram Voice Agent Platform, combining Deepgram's models with Pipecat's orchestration layer, alongside two new models optimized for live conversation and a new Global API.
Why Real-Time Voice Infrastructure Matters
Voice AI systems require tightly orchestrated, low-latency pipelines that handle turn-taking, interruption detection, and packet routing in real time. Building production-grade agents has traditionally required stitching together distinct systems for audio transport, inference, and conversational state.
The voice sector has seen substantial capital deployment across both model creators and infrastructure specialists. Model developers like ElevenLabs, Gradium, and Fish Audio have closed multi-million-dollar rounds, while infrastructure platforms such as LiveKit and Vapi have also raised significant venture capital, according to TechCrunch.
By securing the 18 engineers who built the Pipecat framework, Deepgram positions itself to deliver an end-to-end platform spanning from foundational speech models to global packet routing. Deepgram reports that over 650,000 developers and 5,000 enterprise organizations use its tools to process conversational audio. Today, Deepgram's customers deploy AI agents at production scale for customer support, robotics, and industrial embedded devices.
Questions people ask
What happens to the open-source Pipecat framework after the acquisition?
Pipecat will remain free and open source under its current BSD V2 license. It will be governed by a newly created independent Pipecat Foundation with a steering committee that includes industry participants and Deepgram competitors.
What is the Real-Time Context Standard (RCS)?
RCS is an open protocol co-developed by Deepgram and Pipecat. It transmits rich audio metadata—such as tone, background noise, or whispered speech—alongside transcribed text to language models, allowing voice agents to adjust their responses accordingly.
Will existing Daily customers lose access to their services?
No. Deepgram announced that it will continue to sell, support, and actively develop all of Daily's existing commercial products, including its WebRTC and telephony APIs.
Sources
- Deepgram acquires Pipecat-maker Daily to bolster its voice AI infrastructureTechCrunch, Oct 29, 2026
- Deepgram is acquiring Daily, makers of Pipecat, to offer the first complete, production-ready, open stack for conversational voice AIDeepgram, Oct 29, 2026
- Deepgram Is Acquiring Daily and Establishing an Independent Pipecat FoundationDaily API: Developer Tips to Build Real-time Voice, Video, and AI into Apps, Oct 29, 2026
How this story was made: the newsroom picked it up from Techmeme and Google News, gathered the full text of the sources above, and drafted it with AI assistance. Every factual claim was then checked against those sources before publishing (22 claims checked). Illustrations marked as AI-generated are not photographs. Spotted an error? Tell us.
Published November 2, 2026 at 01:10 UTC
