Understanding of games through their real-life performance is customarily done by overall metrics such as install-rates, or though solutions tailored to a specific game. At Sett, we're building a revolutionary, systematic yet adaptive solution, a unified solution that nevertheless analyzes each game appropriately according to its own universe.

Gameplaying agents have a history as old as AI itself. Modern agents can play on par with humans or even beat them. But playing as humans is a wholly different matter. At Sett, we are training a foundational model that captures and reflects the many aspects which influence human gameplay, in order to guide the game generation and optimize for fun.

Completing game details and even implementing them is easy for LLMs - as long as they are familiar enough. But step outside the tired templates, and things quickly break down. At Sett, building on our game creation ability, we instill true game design intelligence in order to produce refined yet novel creations. The process is managed in a structured way that allows for efficient exploration of the full space of potentialities - quality games and them alone.