- Founded by two leading robotics and AI researchers at INRIA and a manufacturing expert who led multiple teams at Renault
- The software’s abductive learning model enables humanoid robots to learn new industrial tasks with only ten demonstrations, significantly shortening deployment time
- The funding will support the expansion of its research and robotics teams and the deployment of its first industrial robot-learning models
- Initial deployments will take place at major manufacturers in the automotive and consumer goods sector
Bleu Robotics, the Paris-based startup building AI models for industrial humanoid robots, has raised a €5 million pre-seed round, co-led by Contrarian Ventures and OVNI Capital, with participation from Kvanted, WEPA Ventures, Motier Ventures and Charlie Songhurst, technology investor and board member at Meta. The company will use the funds to grow its team and scale its software solution across industries in pursuit of its goal to reduce the full humanoid robot deployment process to one hour or shorter.
According to the International Federation of Robotics, there were 5 million industrial robots operating globally in 2025. However, these robots are built to repeat fixed tasks, and are unable to adapt to the high-mix, low-volume production now common in modern manufacturing, so that work still falls to human workers.
Humanoid robots are needed to automate a wider range of complex tasks. While hardware has advanced significantly in recent years, software development has struggled to keep pace, limiting real-world deployment. Interact Analysis found that of the 20,000-plus humanoids produced last year, only 10% reached real-world applications. Most of these robots today rely on end-to-end models, which learn everything in a single black box and require massive amounts of data to achieve production-ready reliability. As a result, deployment remains costly and time-consuming for manufacturers, often requiring hundreds of hours of training data, with no guarantee the robot will perform reliably.
Bleu takes a different approach to close the gap between hardware and software. Its technology removes the need for those large task-specific datasets and long training cycles, allowing manufacturers to deploy robots cost-effectively and at scale. Instead of an end-to-end model, Bleu Robotics’ abductive learning model allows the robot to learn a new task from as few as ten operator demonstrations, while keeping the explainability required for industrial safety and the 99%+ accuracy that production lines demand.
Also Read: Aetina Introduces AIE-KT78/68 for Integrated AI Perception, Decision-Making and Control in Robotics
The model eliminates the need for vast amounts of training data by inferring the most likely intent behind an operator’s demonstration, including the task’s goal and constraints, and determining how to reproduce it reliably. The software can be plugged into any humanoid platform. Operators teach it new tasks directly at the workstation using Bleu’s training devices, handheld grippers that remove the need for coding, and can retrain the same robot as production needs change or move it to another site.
Benoît Berkoukchi, co-founder of Bleu Robotics, said, “Humanoid robots in manufacturing are the next automotive sector, with the potential to help Europe and the US reshore production. But most robotics companies are trying to train humanoids with millions of data points and build fully general-purpose models. That approach isn’t suited to manufacturing. There isn’t enough data on factory work, and there won’t be internet-scale data, the kind LLMs were built on, for robots for a long time. Our methodology means we can do more with less.”
Thomas Poitrineau, Investor at Contrarian Ventures, said, “We were attracted to Bleu Robotics because of their ambitious choice to rethink how humanoid robotics software should work from first principles. It’s rare to find a company led by two world-class robotics researchers and a manufacturing expert, which is also why their software works differently from their competition. Their goal is far bigger than simply making the best model in a lab. It’s an approach that can already scale, putting robots to work in factories within months rather than years.”
Bleu Robotics was founded in Paris in December 2025 by Benoît Berkoukchi, Jean-Baptiste Mouret and Serena Ivaldi. Berkoukchi spent years managing large teams on automotive assembly lines at Renault. Mouret and Ivaldi are research directors at INRIA, having spent the last 15 years as leading academics in the AI and robotics space. They have published hundreds of scientific papers in prestigious journals, including in Nature, and organized some of the world’s leading humanoid robotics conferences.
Bleu is already working with two major industrial companies while building partnerships with robot distributors. With each deployment, the model will benefit from real-world data that makes it more reliable and faster to roll out. The ambition is to become the software of choice for manufacturers to deploy and manage robotic work at scale.
