Research project / 01
Project One
Reinforcement learning for robotics, games and decision-making chatbots.
Introduction
A different dimension of decision making.
Project One is a state-of-the-art research project where we explore reinforcement learning for robotics, games and decision-making chatbots—a different dimension of Rasa.
The project is inspired by TypeSafe AI and its demonstration that decision-making models could offer an alternative future to expensive and time-consuming language models.
Project One is still at an early stage. We plan to invest significant resources and build a dedicated team to create datasets, practical examples and open-source models.
Collaboration
Work that changes the field.
We are open to collaborating with ambitious teams and with existing teams inside Sleeping AI. For us, collaboration means making a significant contribution to the field—work capable of radically changing it.
Start a conversationInteractive demo / System One
Chess One
We made Chess One to demonstrate where System One models fit naturally and how we approach this field. Make a move and have fun.
Optional access
Play the full match.
Play six moves with our demo key, or use your own TypeSafe API key to finish the match and start new games.
Kept only in this tab’s memory. Never saved by this site.