French robotics-focused startup Genesis AI has released its first foundational model, GENE-26.5, alongside a human-like hand built in-house that can perform complex tasks such as cooking, solving a Rubik’s Cube, and playing the piano. The announcement marks a significant step for the company, which has positioned itself at the intersection of artificial intelligence and physical robotics.
The company, which completed a $105 million seed funding round in July of last year, began with the goal of building an AI model to power real-world robots. However, co-founder and chief executive Zhou Xian explained that the team soon realized that relying on third-party hardware limited their ability to optimize the entire system. That realization prompted Genesis AI to design and manufacture its own robotic hand, giving the company precise control over the full stack—from the model's decision-making logic to the fine motor movements of the end effector.
A Hand That Mimics Human Anatomy
Most robotic hands developed by other companies feature two or three fingers, designed primarily for simple gripping tasks. The Genesis AI hand, by contrast, closely resembles a human appendage, with five fingers and a structure that mirrors natural biomechanics. This design choice is not merely aesthetic; it allows data collected from human demonstrations to be transferred directly to the robotic device without significant adjustments. The closer the physical form is to a human hand, the more directly the model can interpret actions such as grasping a delicate object, rotating a wrist, or applying precise fingertip pressure.
The hand’s dexterity was showcased in a demonstration video released by the company. In the footage, a pair of Genesis AI hands work together to prepare a smoothie, carefully selecting fruits, placing them into a blender, and activating the machine. Another segment shows a single hand playing a keyboard with notable speed and accuracy. These tasks require not only fine motor control but also the ability to process sensory feedback and adjust movements in real time—capabilities that have long been challenging for robotics developers.
Data Collection Through Sensor Gloves
In parallel with the robotic hand, Genesis AI has developed lightweight sensor gloves designed to be worn by workers in fields such as pharmaceuticals and manufacturing. These gloves capture detailed motion data as workers perform intricate tasks—measuring finger angles, grip strength, and subtle hand movements—and transmit that data to the AI model. The company said that data from the gloves can be combined with information gleaned from videos of workers carrying out tasks, creating a richer dataset for training the model to understand human manipulation.
The concept raises practical questions, however. Workers may be understandably hesitant to help train a robotic system that could eventually replace them. Genesis AI has not publicly addressed these concerns in detail, but the company noted that the gloves are lightweight and non-intrusive, designed to be worn over existing work attire without hindering movement. The pharmaceutical and manufacturing sectors, both of which face labor shortages and rising costs, could be early adopters of the technology—provided they can navigate the broader workforce implications.
Foundational Model GENE-26.5
The foundational model at the heart of Genesis AI’s system, GENE-26.5, is designed to power not just the company’s own hardware but potentially other robots as well. Foundational models in robotics are trained on vast amounts of data to develop a general understanding of physical tasks, rather than being specialized for a single action. The “26.5” likely refers to the model's parameter count or version number, though the company has not disclosed full technical specifications. Zhou Xian emphasized that the model’s purpose is to enable real-world robots to perceive, reason, and act in unstructured environments—far beyond the controlled settings of traditional industrial automation.
The decision to build the hand in-house was driven by the need for accurate control over the full stack. By owning both the model and the hardware, Genesis AI can ensure that the model’s outputs translate into precise motor commands, with no compatibility issues or performance bottlenecks. This vertical integration approach is reminiscent of strategies used by major AI companies that build custom silicon to optimize their software. For a startup, however, it is an ambitious undertaking that requires deep expertise in both algorithms and mechatronics.
Background and Founding Team
Genesis AI was founded by Zhou Xian and Theophile Gervet, who serve as CEO and president, respectively. Gervet previously worked at France’s Mistral AI, a leading European artificial intelligence company known for its large language models. The connection to Mistral reflects the broader European AI ecosystem, which has been gaining momentum with significant investment and a growing pool of talent.
The company maintains a substantial presence in Europe, which it attributes to the continent’s large talent base and the potential market of industrial customers. Europe is home to numerous manufacturing and pharmaceutical companies that could benefit from advanced robotics, and the region has been proactive in setting standards for AI and automation. Genesis AI’s team of 60 people is split roughly evenly between the United States and Europe, with a slightly larger group based in San Carlos, California. The company also has offices in Paris and London, giving it a foothold in both major innovation hubs.
Future: Full-Body Robot
Looking ahead, Genesis AI has revealed plans to supplement its mechanical hand with a full-body, general-purpose robot. While the hand alone is capable of performing tabletop tasks, many real-world applications require mobility, arm movement, and the ability to navigate environments. A full-body robot would combine the dexterity of the current hand with torso, arm, and wheeled or legged locomotion. The company has not provided a timeline for this next phase, but the announcement suggests that Genesis AI intends to evolve from a component supplier into a complete robotics platform.
The development of a full-body robot would place Genesis AI in direct competition with established players in the humanoid robotics space, such as Boston Dynamics, Figure AI, and Tesla’s Optimus. However, the company’s focus on a foundational model that can be continuously improved with real-world data may offer a distinct advantage. Rather than hard-coding specific movements, GENE-26.5 learns from demonstrations and feedback, which could allow the robot to adapt to new tasks quickly.
Investors and Industry Impact
The company’s seed funding round in July was co-led by Eclipse and Khosla Ventures, with additional backing from Bpifrance, HSG, former Google chief Eric Schmidt, telecoms entrepreneur Xavier Niel, MIT Computer Science and Artificial Intelligence Laboratory director Daniela Rus, and Apple distinguished scientist Vladlen Koltun. This distinguished group of investors reflects strong confidence in Genesis AI’s vision. The involvement of figures like Schmidt and Rus underscores the potential of the technology to reshape industries ranging from logistics to healthcare.
The robotics AI sector has seen a surge in investment over the past few years, driven by advances in large language models and reinforcement learning. Companies are racing to build general-purpose robots that can perform tasks with human-level dexterity, which could address labor shortages and improve safety in hazardous environments. Genesis AI’s approach—developing a foundational model and then building hardware to match—is a unique angle that could accelerate progress in the field.
However, challenges remain. The complexity of human hand manipulation is immense, requiring the synergy of touch sensors, force control, visual perception, and motor coordination. While Genesis AI’s demonstration video is impressive, it is unclear how well the system performs in less controlled conditions over long periods. Reliability, durability, and cost are critical factors that will determine whether the technology can move from the lab to commercial deployment.
The sensor gloves, too, present a potential avenue for growth. By capturing data from human workers, Genesis AI could build a dataset far larger than what is possible through manual robotics demonstrations. This data could be used to train the model on a wide variety of tasks, from assembling electronic components to handling fragile laboratory equipment. The gloves could also be sold as a separate product, generating revenue while the company continues to develop its full-body robot.
Genesis AI’s strategic focus on Europe is notable in an industry dominated by American and Chinese companies. With offices in Paris and London, the startup is well-positioned to tap into European talent and industrial clients. France, in particular, has been investing heavily in artificial intelligence infrastructure, and the French government has expressed strong support for AI startups. The company’s French roots, combined with its California presence, give it a transatlantic identity that may prove advantageous in navigating international regulations.
As Genesis AI works toward its goal of a full-body robot, the implications for the broader economy are profound. If the technology succeeds, it could transform manufacturing lines, warehouses, and even homes, enabling automation of tasks that were previously impossible to mechanize. The ability to transfer human skills to a robotic system—through video observation and wearable sensors—could also shorten the time it takes to train robots for new roles. This would be a significant departure from traditional programming, where every action must be explicitly coded.
The company’s model-first approach is part of a larger trend in robotics, where the boundary between software and hardware is blurring. Just as large language models like GPT have become the foundation for many AI applications, a general-purpose robotics model could serve as the “brain” for a wide array of machines. GENE-26.5 is an early attempt at this vision, and the company’s willingness to build its own hardware suggests that it sees the integration of sensing and actuation as a core part of the recipe for success.
The road ahead will not be easy. Genesis AI faces competition from well-funded startups and tech giants, each with their own approach to embodied AI. Some are focusing on bipedal locomotion, others on high-speed manipulation, and still others on autonomous navigation. Genesis AI’s bet is that a human-like hand paired with a powerful predictive model will create a data flywheel: more demonstrations lead to a better model, which enables more advanced hardware, which in turn attracts more customers and investors.
For now, the company is celebrating its first major product launch. The combination of a new foundational model and a human-like robotic hand is a bold statement of intent. It signals that Genesis AI is not content to be an also-ran in the robotics race. Instead, it aims to define the standard for dexterous manipulation in artificial intelligence. Whether the hand can move from viral demonstration videos to factory floors and hospital wards remains to be seen, but the foundation has been laid.
With a team of 60 people spread across California, Paris, and London, and a growing list of influential backers, Genesis AI is poised to play a significant role in the next chapter of robotics. The company’s progress will be closely watched by researchers, investors, and industry leaders alike, as it attempts to bridge the gap between human capability and machine execution.
Source: Silicon UK News