Deepgram Acquires Daily to Expand Voice AI Capabilities
Deepgram integrates Pipecat framework while establishing an independent foundation for open-source voice development.
1Strategic Acquisition Details
Deepgram has acquired Daily, the developer behind the Pipecat open-source framework for voice agents. The deal, announced at the Deepgram Speak conference in San Francisco, aims to unify voice and video streaming capabilities. While financial terms remain undisclosed, the integration marks a significant consolidation within the real-time voice AI infrastructure market.
2Open Source Commitment
Pipecat will maintain its BSD V2 license, ensuring it remains free and open source. To guarantee vendor neutrality, the companies are establishing an independent Pipecat Foundation. This structure allows competitors to participate in governance, a condition Daily co-founder Kwindla Hultman Kramer prioritized during the acquisition negotiations with Deepgram CEO Scott Stephenson.
3Technical Framework Capabilities
Pipecat serves as the connective tissue for voice agents, linking speech-to-text, large language models, and text-to-speech components. The framework currently boasts nearly 400 contributors and over 200 integrations. Developed by a team of 18 engineers, it is designed to manage complex voice and video streams while maintaining minimal latency for users.
4Infrastructure and Scale
Daily operates a network of 75 points of presence to support real-time voice and video infrastructure. The company previously raised over $62 million from investors like Tiger Global and Nat Friedman. Deepgram, now a unicorn following a $130 million Series C round, supports over 5,000 organizations running its speech models in production.
5New Voice Agent Platform
Deepgram is integrating Pipecat as the core of its new Voice Agent Platform. Chief product officer Ed Anuff is previewing this platform alongside two new models specifically designed for live conversation. The company will continue to sell and support all existing Daily products while transitioning the team into the Deepgram organization.
6Real-Time Context Standard
Deepgram introduced the Real-Time Context Standard (RCS) to address information loss in current voice systems. By adding metadata regarding audio signals, model instructions, and reply guidance, RCS allows agents to interpret nuances like tone or background noise. This protocol enables more sophisticated interactions, such as whispering back or requesting clarification.
7Long-term Vendor Neutrality
CEO Scott Stephenson emphasized that Deepgram intends to remain vendor-neutral, encouraging developers to select the best models for their specific use cases. This approach contrasts with closed ecosystems, aiming to foster broader adoption of voice AI. The company has maintained a partnership with Daily since 2019, utilizing Deepgram for transcription services.
8Market Consolidation Trends
The acquisition reflects a broader trend of significant investment in voice AI infrastructure and model development throughout this year. Other recent market activity includes Vapi’s Series B funding, Fish Audio’s seed round, and ElevenLabs' tender offer discussions. These moves highlight the intense competition to build the foundational layers for future voice-based applications.
Original article: The Next Web — Deepgram Acquires Daily to Expand Voice AI Capabilities