The frontier AI reasoning lab
Building the foundations for reasoning over large-scale tabular data and AI to improve how humans and agents make decisions.
Our technology
Frontier tabular models
Develop high performance tabular AI models to transform enterprise decisioning and process automation, including foundation models and transformer-based time series models.
Interpretability
Build efficient algorithms for interpreting tabular AI models and extract critical context to unlock actionable decisioning and accelerate automation for both traditional and frontier models.
Verifiability
Build trust into tabular model intelligence with new methods to rigorously check that interpretations correctly capture model behavior and enable realistic recourse.
Blog
Updates and insights from the OuterProduct team
Fundamental AI research
From our founders' labs at MIT and UCSD
Science · February 19, 2026Steering and Monitoring of AI Models (opens in new tab)Daniel Beaglehole, Adit Radhakrishnan, Enric Boix-Adserà, Mikhail BelkinRead the paperScience · March 7, 2024Feature Learning in Neural Networks (opens in new tab)Adit Radhakrishnan, Daniel Beaglehole, Parthe Pandit, Mikhail BelkinRead the paper
