Fundamental

Fundamental is an AI company building general-purpose foundation models for structured data. Its focus is Large Tabular Models (LTMs): models designed for the rows and columns of enterprise databases, the records on which business decisions are made. The company was founded and is led by DeepMind alumni, and its research agenda covers meta-learning, architecture research, generative modelling, in-context learning and large-scale ML systems, alongside code generation and structured data processing.

The company's product, NEXUS, is billed as the world's most powerful Large Tabular Model. It is purpose-built for structured records and adapts to novel tasks through in-context learning, drawing on shared experience across billions of examples. Fundamental states that its models are trusted by Fortune 100 companies, operating in a domain where trillions of dollars of business decisions each year rest on tabular data.

The work sits at the intersection of research and production systems, with an emphasis on fundamental machine-learning problems rather than narrow applications. Fundamental describes its mission as creating a world of abundance by giving humanity the power to predict, and is backed by world-class investors. The company operates in enterprise software, enterprise decision-making and structured-data analytics.