French robotics-focused start-up Genesis AI has released its first foundational model, GENE-26.5, along with a human-like hand built in-house that is capable of carrying out complex tasks such as cooking or solving a Rubik’s Cube. The company, which completed a $105 million seed funding round in July of last year, began with a focus on a model to power real-world robotics, then built its own hand in order to have more accurate control over the full stack, according to co-founder and chief executive Zhou Xian.
The hand developed by Genesis AI differs markedly from other robotic hands in the market. While many competitors produce two- or three-fingered grippers for simple pick-and-place operations, the unit from Genesis AI closely resembles a human appendage, complete with five fingers and joints that mimic human anatomy. This design allows data collected from human demonstrations to be more directly transferred to the device, reducing the gap between simulation and reality. In a demonstration video, Genesis showed a pair of its hands carrying out intricate processes such as preparing a smoothie, playing the piano, or solving a Rubik’s Cube with dexterity that approaches human ability.
The Genesis Hand: Engineering and Capabilities
Building a robotic hand that can manipulate objects with the same versatility as a human is one of the grand challenges in robotics. Genesis AI’s hand uses advanced actuators and sensors to achieve fine motor control. Each finger has multiple degrees of freedom, enabling tasks that require precision, such as handling fragile glassware or assembling complex components. The hand is also designed to be durable and low-cost enough for mass deployment in industrial settings.
Genesis AI’s approach stands out because it combines a custom hardware design with a powerful AI model. The foundational model, GENE-26.5, is a neural network trained on extensive datasets of human manipulation. By using a human-like hand, the company ensures that the training data—collected from people wearing sensor gloves—can be directly applied without the need for complex mapping algorithms. This end-to-end integration accelerates the learning process and improves generalization to new tasks.
Foundational Model GENE-26.5
GENE-26.5 is the first model released by Genesis AI, and it serves as the brain behind the robotic hand. It is a large-scale transformer-based model trained on diverse manipulation data, including video recordings and glove-captured motion. The model can generate low-level motor commands for the hand to perform a wide array of tasks, from household chores to precision assembly. The name “GENE-26.5” reflects the model size and version, hinting at 26.5 billion parameters, though the company has not officially confirmed the exact size.
Training such a model requires massive computational resources and high-quality data. Genesis AI has built a proprietary data pipeline that allows it to collect real-world manipulation data efficiently. The combination of video and glove data enables the model to learn both visual cues—such as the position of objects—and tactile feedback, like pressure and grip strength. This multi-modal approach improves the model’s ability to handle unexpected situations, such as objects with irregular shapes or varying textures.
Sensor Gloves and Data Collection
To gather training data, Genesis AI has created lightweight sensor gloves that can be worn by workers in fields such as pharmaceuticals or manufacturing. These gloves are equipped with sensors that track hand movements, finger positions, and force application. When a worker performs a task like assembling a smartphone or pouring a chemical, the gloves capture every nuance of the manipulation. This data is then used to train the GENE-26.5 model, enabling the robotic hand to replicate the actions with high fidelity.
However, the use of worker-derived data raises ethical and practical questions. While the gloves are non-invasive and designed for comfort, some workers may be reluctant to help train a potential robotic replacement. Genesis AI has addressed this by emphasizing that the technology is meant to augment human capabilities rather than replace them. In many settings, robots can handle repetitive or hazardous tasks, freeing humans for higher-level decision-making. Nevertheless, the societal impact of such automation remains a topic of debate.
Data from the gloves can also be combined with information gleaned from existing videos of workers carrying out tasks. This fusion of modalities reduces the need for dedicated data collection sessions and allows the AI to learn from a vast library of human demonstrations available online. By leveraging both structured and unstructured data, Genesis AI accelerates the development of robust manipulation skills.
The Road to a Full-Body Robot
Genesis AI has ambitious plans beyond just a hand. The company stated that it is working to supplement the mechanical hand with a full-body, general-purpose robot. Such a robot would integrate the dexterous hand with a mobile platform, arms, and a sensor suite, allowing it to operate in human environments. This vision aligns with the broader industry goal of creating versatile humanoid robots that can perform a wide range of tasks, from warehouse logistics to elderly care.
The development of a full-body robot presents significant engineering challenges: balance, locomotion, power management, and safe human-robot interaction. Genesis AI’s focus on a human-like hand indicates that the company believes manipulation is the most critical capability to solve first. Once the hand can reliably perform complex tasks, the chassis can be designed around it. The company’s seed funding of $105 million provides ample resources to pursue this long-term goal. The funding round was co-led by Eclipse and Khosla Ventures, with participation from Bpifrance, HSG, and prominent individuals such as former Google chief Eric Schmidt, telecom entrepreneur Xavier Niel, MIT CSAIL director Daniela Rus, and Apple distinguished scientist Vladlen Koltun.
Team and Operations
Genesis AI’s team of 60 people is split roughly evenly between the US and Europe, with a slightly larger group based in San Carlos, California. The company also maintains offices in Paris and London. Co-founder and president Theophile Gervet previously worked at France’s Mistral AI, a leading European AI company. The European presence is strategic: the continent has a large talent base in robotics and AI, as well as a potential market of industrial customers in manufacturing, automotive, and pharma. The company leverages academic partnerships and access to European research institutions to stay at the cutting edge.
The leadership team combines deep expertise in AI, hardware design, and business scaling. Zhou Xian, the CEO, has a background in robotics and entrepreneurship, while Theophile Gervet brings experience from one of Europe’s most prominent AI startups. This blend positions Genesis AI to bridge the gap between research breakthroughs and commercial products.
Implications for Industry and Society
The release of GENE-26.5 and the robotic hand marks a significant step toward practical robotic manipulation. Industrial sectors that rely on manual dexterity—such as electronics assembly, food preparation, and laboratory analysis—could benefit greatly. By automating delicate tasks, companies can improve consistency, reduce errors, and mitigate labor shortages. Moreover, the sensor glove technology could be used in training and ergonomics, helping workers perform tasks more efficiently or with less risk of injury.
However, the road to widespread adoption is long. Cost, reliability, and safety remain major hurdles. Robotic hands must operate reliably for thousands of hours without failure, and they must be safe to work alongside humans. Genesis AI will need to demonstrate that its system can handle real-world variability and pass stringent regulatory standards, particularly in industries like pharmaceuticals where contamination is a concern.
From a broader perspective, the development of humanoid robots capable of general manipulation could reshape the labor market. While some jobs may be automated, new roles will likely emerge in robot maintenance, programming, and oversight. The technology also raises questions about income inequality and the need for retraining programs. Genesis AI’s approach of augmenting rather than replacing human workers may mitigate some of these concerns, but the debate will continue as the technology matures.
In the meantime, the company is focused on refining its hand and model, expanding its data collection efforts, and working toward that full-body robot. The combination of a sophisticated AI model and a human-like hand gives Genesis AI a unique advantage in solving the manipulation problem. As the company grows its team and continues to iterate on its designs, it could become a key player in the next wave of robotics innovation.
Source: Silicon UK News