Generalist AI models that interface universally across embodiments & physical devices
Subscribe for research and release updates!
Intelligence is an interfacing problem
And in interfacing lies the solution: human-generated data, synthetic data, heterogenous simulations and feedback from physical environments, using cross-domain insights from ML, robotics and first principles.
Automated Deep Tech
The loop between knowledge acquisition, verification and application needs to be tighter, faster, and increasingly off-loaded to artificial intelligence. That's why we're automating our aggressively hands-on deep tech research & development.
RND
Automated research and development pipeline with autonomous agents harvesting knowledge alongside our team to bridge theory and practice from first principles.
Universal Interfacing Intelligence
Generality is achieved once a system can autonomously adapt and continuously learn by interfacing with both the digital and physical, granting it sufficient causal efficacy and understanding to self-generate solutions and architectural adjustments upon failure.
AGI
Our central general-purpose AI model effort focused on adaptable general intelligence with autonomous learning to power real-world self-generating systems.
GYM
Foundational training suite and approach, aimed at developing generalizable skills.
SIM
Simulation sandbox suite that combines heterogeneous data sources to prep models for the real world.
New Engineering Paradigm
Most of our engineering is by humans and for us humans, which comes with a lot of design drawbacks and unnecessary abstraction layers. We are reinventing how to think about tech stacks and engineering, anticipating humans not needing to be in the loop.
ROB
Robotics initiative leveraging SIM and AGI, combining rapid prototyping, morphological computation, and intelligence for real-world interfacing to achieve universal embodiment.
Jobs
Join in on the magic of thinking up new ways to build AI & Technology
FAQ
A few questions we can already answer.