Deepgram Buys Daily, Leaves Pipecat Open

Deepgram said on 29 October that it is acquiring Daily, the company behind Pipecat, an open-source framework used to build real-time voice and video AI systems. The companies did not disclose the purchase price, but both said Pipecat will remain open-source.

That promise matters because Pipecat sits in a part of the AI stack that users rarely see and developers feel immediately: the machinery that moves audio and video with as little delay as possible while connecting speech recognition, language models and synthetic speech. If these systems hesitate, interrupt badly or lose context, the problem is often infrastructure before it is intelligence.

Daily built Pipecat while also selling hosted voice and video infrastructure. Deepgram, best known for speech technology, appears to be buying more than a product. It is buying a closer position in the plumbing of live AI interaction, where speed, orchestration and compatibility can matter as much as model quality.

According to TechCrunch, Daily co-founder Kwindla Kramer said the company had received acquisition interest but chose Deepgram in part because of a commitment to keep Pipecat open-source and vendor-neutral. Deepgram CEO Scott Stephenson said the company wanted to help developers mix and match models rather than force a single closed route through its own tools.

That is the stated strategy. What is established is narrower: Deepgram is taking over Daily; Pipecat is supposed to stay open-source; and Deepgram is also introducing a Real-Time Context Standard, an open protocol developed with the Pipecat team for passing signals such as tone or whispered speech through a voice agent pipeline.

What remains unknown is how independent Pipecat will feel in practice once it sits inside a much larger commercial player. Open-source projects can remain legally open while drifting toward one company’s priorities, support structure or preferred integrations. Deepgram is arguing the opposite case: that sitting in the middle of a fast-growing ecosystem is commercially useful precisely because it stays open.

The acquisition lands in a voice AI market that is thick with money and increasingly vertical ambition. Companies are no longer competing only on the quality of transcription or synthetic voices. They are trying to control the layer that lets those parts work together in real time, where a delay of a fraction of a second can make software feel less like a tool and more like a dropped call.