Products
Data
Foundation models of human behavior are bottlenecked on data. The public record is full of what people said and thin on what people did.
We source behavioral data, structure it into sequences, and link every interaction to real content, so a model can learn from it.
Models
Sequence models trained on that data. Given what a person has done, they predict what that person does next.
Recommendation first. Simulation, cognition, and domain models to follow, as heads on one backbone, working toward a single model of what a human would do in any situation.
Applications
What runs on the model. Each is an adaptation of the same backbone rather than a separate system.
- Recommendation
- Matching
- Personalization
- Memory
- Simulation
- Forecasting