Our research direction is Omnibody Intelligence: shared intelligence powered by world models and designed to adapt across different bodies, tasks, and environments.
01 / The world
Intelligence begins in the world.
A digital system can reason over a representation it is given. An intelligent physical system must also form and update representations while the world changes around it. Time passes, observations are incomplete, and every action can alter what happens next.
This makes uncertainty part of the problem, not an edge case. HumanXAI is researching world-model-based systems designed to identify relevant structure, retain context, anticipate possible consequences, and support planning within defined boundaries.
02 / Memory
What came before changes what to do next.
Immediate reactions are not enough for many physical tasks. Useful decisions can depend on events that occurred seconds, hours, or much earlier. Persistent memory can help an intelligent system connect those events without treating every observation as an isolated moment.
Our public research direction treats memory as working context: information that can inform understanding, reasoning, and planning over time. The aim is to preserve what matters while remaining responsive to a changing environment.
03 / Embodiment
Shared knowledge. Different bodies.
Robotic bodies differ in shape, sensing, actuation, and physical constraints. HumanXAI is investigating shared representations intended to carry useful knowledge across those differences without assuming that every body behaves in the same way.
This is a research objective, not a claim of proven universal transfer. Each embodiment still requires body-specific control and careful validation. The thesis is that useful intelligence should not always need to be rebuilt from the beginning when the body changes.
04 / Reliability
Action must account for consequence.
Physical autonomy operates where mistakes have real effects. Systems must reason about uncertainty, respect operational limits, and remain controllable as conditions change. Reliability and safety are therefore part of the research problem itself.
HumanXAI is developing this direction for long-term applications in robotics, industrial automation, healthcare, inspection, assistive systems, and field robotics. These are research horizons, not statements of current commercial deployment.
HumanXAI is building an international research and engineering team in Paris and Zürich.
If this vision resonates with you, come build the future with us.
HumanXAI — Intelligence for the physical world.