AssemblyAI, founded in 2017, provides Speech AI models and an API for transcribing and understanding human speech at scale. The company serves 600 million inference calls monthly, processes over 1 million hours of audio daily, and powers more than 2 billion end-user experiences across production applications.
The core product is an API that companies integrate into their platforms to add speech transcription and understanding capabilities. Notable customers include Zoom, Granola, Fireflies, Cluely, and Calabrio. AssemblyAI's Universal series of models form the technical foundation, with variants optimised for streaming inference and features such as keyterm prompting and multilingual support. The company has raised over $115 million from investors including Accel, Insight Partners, Y Combinator's AI Fund, and individuals such as Patrick and John Collison, Nat Friedman, and Daniel Gross.
AssemblyAI operates across several industry verticals: voice agents, meeting assistants, contact centres, medical scribing and healthcare documentation, and video conferencing platforms. The company's technical focus centres on reducing latency and hallucinations while maintaining accuracy in speech recognition - addressing what it characterises as substantial gaps in a tens-of-billions-of-dollars market.