Deepgram

Deepgram is a research-driven, foundational AI company that builds voice AI technology and developer tools for instant, accurate, human-like interaction between people and machines. Its work spans speech recognition, speech-to-text, audio processing, foundation models, deep learning, low-latency inference and natural language processing, with an engineering focus on systems that perform reliably at scale.

The company's product is a set of voice-native foundation models aimed at developers building human-to-machine voice applications. Deepgram claims the lowest latency and highest accuracy in voice AI. The company operates on a developer-first basis, supplying tools for engineers rather than end-user applications, and emphasises enterprise-grade reliability in its deployments across voice AI and conversational AI, developer tools, enterprise software and telecommunications.

By its own account, Deepgram has processed more than 50,000 years of audio and over one trillion words. Its technology is used by more than 200,000 developers and over 1,300 organisations. The company raised $130m in a Series C round and is valued at $1.3bn.