The Lausanne-based company has been backed by Look AI Ventures, with the latest funding round led by Faber VC and joined by Techshop Capital, Kickfund, Plug and Play San Francisco, Excellis and Vento. The company did not disclose the size of the round. The new capital will be used to bring Embodied AI’s technology to more customer production lines, increase robot production, improve its data and AI infrastructure, and expand the company’s teams in Lausanne and Rome.
The company is entering a market where industrial automation is already well established, but where many manufacturing tasks remain difficult to automate. Conventional industrial robots are highly effective when they are given a clearly defined, repetitive task in a controlled environment. They can repeatedly move components, weld parts or perform other standardized operations with high precision. The problem comes when production involves frequent changes or when a task requires the kind of fine motor control and adaptability that humans naturally provide. Embodied AI is targeting this part of the manufacturing process, with applications including electronics assembly, cable handling, kitting and other tasks where factories need flexibility rather than simply repetition.
The next challenge for factory automation
For decades, industrial automation has largely been built around the idea of programming machines to perform a predefined sequence of actions. That approach works particularly well in high-volume manufacturing, where the same product can be produced thousands or millions of times and the factory environment remains relatively predictable. However, it becomes much harder to justify when production involves frequent changeovers, different components or products, and tasks that require a robot to react to small variations. These are also areas where human workers remain heavily involved, creating an opportunity for a new generation of AI-powered robotics.
Embodied AI believes that advances in artificial intelligence can change this equation by allowing robots to learn from what actually happens on the production line. Instead of trying to build one universal AI model capable of understanding every possible industrial task, the company is developing a platform that collects production data and trains smaller, task-specific models. Several models can work together to handle variations within a particular task, allowing the robot to become more adaptable without requiring the entire system to be rebuilt whenever the production environment changes.
This approach also reflects a broader shift in the robotics industry toward what is increasingly being described as physical AI or embodied AI; AI systems that do not simply generate text, images or predictions, but interact with the physical world. For manufacturers, the promise is significant: instead of simply automating a fixed process, robots could potentially learn how to deal with new situations and improve their performance through experience.
Humans remain part of the loop
Embodied AI is not initially betting on a world where robots operate completely independently. Its system is designed to allow human operators to intervene remotely when a robot encounters a situation it cannot handle. The operator can correct the robot’s behavior while production continues, and the interaction creates additional data that can subsequently be used to improve the underlying AI models.
The company describes this process as an “Embodied AI flywheel”, combining autonomous operation, human intervention, data collection and continuous model improvement. In theory, the more a robot operates in real production environments, the more data it can collect and the better the system can become at dealing with the variations that previously required human workers. Over time, the objective is to reduce the frequency of human intervention and move individual tasks closer to full autonomy.
Safety is another part of the company’s approach. Rather than treating safety purely as a software problem, Embodied AI says its robotic hardware has been engineered from the ground up with physical safety mechanisms incorporated into the design. The company believes that combining purpose-built hardware with AI software and remote teleoperation can help overcome some of the practical barriers that have historically prevented robots from being deployed more broadly in factories.
From the lab to European production lines
The startup is already moving beyond research and development and into commercial manufacturing environments. Embodied AI says it has partnered with global industrial companies to deploy robots in areas including electronics production, kitting and flexible cable handling. It is also working with some of the largest manufacturers in Europe, with applications spanning electronics, logistics and automotive assembly.
That focus on real-world deployment is central to the company’s strategy. Rather than developing a robotics platform first and looking for applications later, Embodied AI is using customer production environments as part of the development process. The company says its immediate priority following the new funding is to bring its technology onto additional customer production lines and demonstrate measurable return on investment. The next challenge will be turning individual deployments into a scalable robotics platform that can be manufactured and deployed across a much larger number of factories.
The company is also connected to some of the major names in the global AI ecosystem. Embodied AI is a Google DeepMind partner and is part of Nvidia Inception, Nvidia’s program supporting startups developing technologies around AI and accelerated computing.
Building a European robotics company
The company’s ambitions extend beyond individual deployments. Embodied AI plans to use part of the new funding to scale its robot production capabilities and improve the infrastructure it uses to collect production data and train its models. It is also planning to expand its research, engineering and commercial teams in Lausanne and Rome, with hiring expected to cover researchers, engineers and commercial leaders.
The company’s founders bring together academic robotics research and experience in building technology companies. Francesco Stella, Embodied AI’s CEO, holds a PhD in Robotics and AI from EPFL, where his thesis received ABB and EPFL awards, and previously conducted research at MIT CSAIL. CTO Kai Junge holds a Master’s degree from Cambridge and a PhD in robotic manipulation from EPFL, while COO Max Polzin trained at ETH Zurich and EPFL and was previously the first employee at Seervision, where he helped lead the company through its exit.
The scientific board includes Professor Cosimo Della Santina of TU Delft and Professor Josie Hughes of EPFL, while the wider team includes scientists and engineers with backgrounds at institutions including Oxford, TU Delft and MIT. The combination of academic robotics expertise and commercial experience is intended to help the company bridge the gap between robotics research and industrial deployment.
For investor Faber VC, that combination is also what makes Embodied AI’s approach relevant to the emerging physical AI market. “By vertically integrating proprietary soft-robotic hardware, teleoperation and autonomy-ready software, Embodied AI is creating a fundamentally new robotics platform, inherently safe, cost-effective and designed for real-world deployment,” said Sofia Santos, Partner at Faber VC.
For Embodied AI, however, the immediate objective is less about building a futuristic general-purpose humanoid robot and more about solving specific problems on today’s factory floors. “Our approach prioritises practicality over the hype of the moment,” said Francesco Stella, Co-founder and CEO of Embodied AI. The company’s new funding will allow it to put that approach to the test at a larger scale, deploying its technology with manufacturers, demonstrating the economic value of its robots and building what it describes as the foundation for a new European robotic workforce.
