French Humanoid Robots Revealed By UMA

UMA, a European Physical AI company, has unveiled the design of its first humanoid robot and introduced a new AI architecture called Real-Time Learning that enables robots to acquire new skills through demonstration rather than manual programming.

The unveiling took place at Machina Summit in Paris, where the company outlined its approach to building robots intended for deployment in factories, warehouses, logistics centres and industrial facilities.

Addressing Labour Shortages

UMA is positioning its humanoid robots as a response to growing pressure on labour markets across advanced economies. The company pointed to research from Korn Ferry suggesting the global economy could face a shortage of 85 million workers by 2030, representing as much as US$8.5 trillion in unrealised economic output.

Rémi Cadène, CEO and Co-Founder of UMA, framed the company’s work in the context of demographic and industrial shifts.

“Demographic, industrial and environmental challenges all point to the same reality: societies need greater productive capacity,” Cadène explained. “We believe intelligent robots will become part of the solution, not as a substitute for people, but as a new class of tools that enables them to devote more time to what machines will never replace: creativity, judgment, innovation and caring for others.”

Designed For Industrial Environments

UMA’s approach centres on building robots capable of working in environments already designed for people, rather than creating technology demonstrations.

The humanoid architecture is intended to allow the robots to use existing tools, integrate into current infrastructure and collaborate alongside human workers without requiring changes to the physical environment.

This engineering approach extends to how the robots learn. By interacting with the physical world in ways that mirror human behaviour, UMA’s robots can learn from demonstrations, receive guidance when needed and improve their performance in operational settings over time.

The ‘Dressed Machine’

UMA has taken a deliberate design direction that differs from other humanoid robotics efforts. Rather than building machines that attempt to appear human or lean heavily into visible mechanical complexity, the company has adopted what it calls the “dressed machine” approach.

The robot combines human-scale proportions with a neutral visor instead of facial features, intended to eliminate ambiguity between person and machine. A soft technical outer shell is paired with intentionally visible mechanical joints, embracing the robot’s identity as a machine rather than concealing it.

The company has stated its objective is not to build a machine that impresses on stage, but one that integrates into industrial operations and becomes a reliable tool over time.

Real-Time Learning

At the core of UMA’s platform is Real-Time Learning, an AI architecture that allows robots to acquire new capabilities from demonstrations, adapt to unfamiliar situations and refine their execution through experience.

The approach draws on the way people learn new tasks – through observation, experimentation, practice and progressive improvement. Rather than requiring engineers to manually reprogram robots for each new application, the platform enables continuous learning from the environment.

Cadène outlined the significance of the approach: the industry’s challenge is no longer building robots capable of executing individual tasks, but building robots capable of learning new ones.

The architecture is intended to make robots more flexible and easier to deploy across a range of industrial settings without extensive reprogramming.

A European Bet On Physical AI

While the humanoid robotics field is largely driven by companies in the United States and China, UMA is positioning Europe as a viable base for the next phase of Physical AI development.

The company argues the region combines scientific research capability, a strong industrial base and growing demand for automation driven by structural labour shortages.

UMA was founded in 2025 by Cadène, Pierre Sermanet, Robert Knight and Simon Alibert. The company is headquartered in Paris, with teams across London and Geneva.

Long Road To Deployment

Cadène acknowledged that large-scale deployment of humanoid robots remains years away.

“We’re still at the beginning of this journey,” he remarked. “Humanoid robots will take years to reach large-scale deployment, just as the internet and smartphones required time before transforming entire industries. We believe intelligent robots will reshape the physical economy in much the same way.”

“One day, robots will contribute to manufacturing future generations of robots, accelerating their own deployment. But our ambition has never been to replace people. It is to expand what people are capable of, to give everyone more time to create, solve problems and focus on what makes us uniquely human. That vision guides every scientific, technological and industrial decision we make at UMA.”

Last Updated on July 7, 2026 by Nick Ross