Fundamental

Fundamental is an AI company building scalable, general-purpose foundation models for tabular and structured enterprise data. Its work sits at the intersection of large tabular models, meta-learning, architecture research, generative modelling and large-scale systems, applied to prediction and decision-making on the structured records held in enterprise databases. The company was founded by DeepMind alumni and is backed by world-class investors; it describes its mission as creating a world of abundance by giving humanity the power to predict.

Its main product, NEXUS, is a Large Tabular Model (LTM) purpose-built for structured and tabular records. Rather than training a separate model for each task, NEXUS leverages shared experience across billions of examples and adapts to novel tasks via in-context learning. Technical domains covered by the team include foundation models, code generation, structured data processing and in-context learning, alongside core machine learning research.

The company targets enterprise data and analytics, where it notes that trillions of dollars of annual business decisions are made on tabular data. Fundamental reports that its models are trusted by Fortune 100 companies. The culture is described as research-oriented, focused on fundamental machine learning challenges.