About
A small, independent AI research and development company working across industries, from games to model development.
The work crosses industries, from games to models to interfaces. What connects it is one question: how do people and intelligent systems work together effectively, safely, naturally and at a cost that makes sense?
We answer it by building. A question becomes an experiment, an experiment becomes a model or a tool, and the result is measured in public. Some of it ships. Some of it stays an experiment, and we say so.
- Open-source frameworks and tools
- Software other people build agents with, starting with Noetic.
- Models
- Small, specialized models, beginning with models that make decisions.
- Benchmarks
- Evaluation suites published alongside the work they measure.
- Research
- Papers and technical reports, each tied to something you can run.
- Interfaces
- New ways for people and models to work on the same thing.
- Games
- AI-native games, which double as our hardest testbed.
We are independent: no outside roadmap and no product quota. We also take on a small amount of applied work with outside teams.