We’re building foundation models for recommendation.

Recommendation systems mediate how billions of people encounter information, ideas, and opportunity. We believe they will become one of the most consequential applications of language models.

Our research brings content, human behavior, and intent into a shared modeling framework. The aim is to build models that reason about preferences, learn from interaction, and put people in control of what they discover.

Our work builds on experience developing language models and recommendation systems at global scale. We’re building toward a future where recommendation is a core capability of intelligence.