In 1984, Susan Kare gave the Macintosh a smile. A small, pixelated face that greeted you like a friend. It was the first time a computer felt human.
We're a research lab building that feeling into AI. Today's AI doesn't interact. Every conversation turns uncanny and spirals. We study how to give AI a continuous sense of what is going on and who it's designed to be, so interactions play by our rules and feel human. Our work spans social psychology, cognitive architectures, world models, and context engineering for AI that interacts, adapts, and grows alongside the people it serves.
| Project | Focus | Type | Status |
|---|---|---|---|
| Manifesto | Interaction world models | Memo | Read ↗ |
| Field | Labeling environment | Notes/Demo | Read ↗ |
| Steering | Lucid LLM interactions | Notes | Read ↗ |
| Injection | Context states @ runtime | Notes | Read ↗ |