French robotics-focused startup Genesis AI has officially entered the artificial intelligence and robotics arena with the release of its first foundational model, GENE-26.5, alongside a remarkably human-like robotic hand built entirely in-house. The hand is capable of performing intricate tasks such as cooking a meal, playing the piano, or solving a Rubik's Cube — feats that demonstrate a level of dexterity rarely achieved by existing robotic systems.
The company, which completed a massive $105 million (£77.2 million) seed funding round in July of the previous year, initially concentrated on developing an AI model to power real-world robotics applications. However, co-founder and chief executive Zhou Xian explained that the team soon realized the need to build their own robotic hardware to gain precise control over the entire technology stack — from the model to the physical actuators.
Unlike robotic hands produced by dozens of other companies, which typically feature only two or three fingers, Genesis AI's device closely mirrors the anatomy and movement of a human hand. This design choice allows data gathered from human demonstrations to be directly transferred to the robot without complex translation or adaptation layers. The hand’s anthropomorphic structure enhances its ability to manipulate objects with a natural grasp, making it suitable for a wide range of industrial and service applications.
Background and Founding Team
Genesis AI was founded by Zhou Xian and Theophile Gervet, both of whom bring deep expertise in artificial intelligence and robotics. Gervet previously worked at Mistral AI, a prominent French AI company known for its large language models. The startup maintains a strong presence in Europe, with offices in Paris and London, in addition to its US headquarters in San Carlos, California. The team, numbering approximately 60 people, is split roughly evenly between the United States and Europe, with a slight concentration on the West Coast.
The seed funding round was co-led by Eclipse and Khosla Ventures, two venture capital firms with extensive portfolios in deep tech. Additional investors included Bpifrance (the French public investment bank), HSG, former Google chief Eric Schmidt, telecom entrepreneur Xavier Niel, MIT Computer Science and Artificial Intelligence Laboratory director Daniela Rus, and Apple distinguished scientist Vladlen Koltun. This high-profile backing underscores the potential that investors see in combining AI with physical robotics.
The GENE-26.5 Model
The GENE-26.5 model is a foundational AI system designed specifically for robotics. It is trained on vast amounts of data, including video demonstrations and sensor recordings, to understand how to control motors, joints, and grippers in real time. The model’s architecture enables generalization across different tasks — meaning that once trained, it can adapt to new manipulation challenges without requiring task-specific retraining. This is a crucial step toward building general-purpose robots that can operate in unstructured environments, such as homes, factories, or laboratories.
Currently, many industrial robots are programmed for single, repetitive tasks. By contrast, Genesis AI aims to create a model that can handle a variety of delicate and complex processes. The company demonstrated this by showing two of its robotic hands working together to prepare a smoothie — a process that involves picking up fruits, opening a blender, pouring liquids, and pressing buttons. The same hands were also shown playing a simplified piano piece, pressing keys with appropriate force and timing.
Sensor Gloves and Data Collection
In addition to the robotic hand, Genesis AI has created lightweight sensor gloves that can be worn by human workers in fields such as pharmaceuticals, electronics assembly, or manufacturing. These gloves capture detailed motion and force data as the wearer performs tasks. That data is then fed into the AI model to teach it how to replicate those movements with the robotic hand. The system can also combine glove data with information extracted from standard video recordings of workers, reducing the need for expensive instrumentation.
However, the company acknowledges that the use of such technology raises questions about workforce dynamics. Workers might be hesitant to help train a potential robotic replacement, especially if automation could lead to job displacement. Genesis AI has not publicly addressed how it plans to manage these ethical and social concerns, but the sensor gloves represent a potentially valuable tool for non-disruptive data collection if implemented with proper worker consent and compensation.
Future Plans: Full-Body Robot
Genesis AI is already working on extending the robotic hand into a complete humanoid or full-body general-purpose robot. The company’s long-term vision is to create a versatile machine that can operate in environments designed for humans, using the same tools and interfaces. This would enable deployment in logistics, healthcare, home assistance, and manufacturing without requiring significant infrastructure changes.
The development of a full-body robot poses substantial engineering challenges. It requires robust balance control, locomotion, and coordination between the arms, hands, and visual perception systems. Many companies, such as Tesla (with Optimus), Boston Dynamics, and Figure AI, are pursuing similar goals. However, Genesis AI’s focus on a highly dexterous hand gives it a distinctive advantage in manipulation tasks, which are often the bottleneck for humanoid robots.
Industry Context and Competition
The robotics and AI market is experiencing a renaissance, fueled by advances in large language models, computer vision, and hardware miniaturization. A growing number of startups and established technology firms are racing to build robots that can perform useful work in dynamic environments. Genesis AI’s $105 million seed round is among the largest in the industry, indicating strong investor belief in the company’s approach.
The field of robotic hands is particularly crowded, with products like the Shadow Dexterous Hand, the Robotiq gripper, and various designs from research institutions. Yet most of these are either underactuated or designed for specific tasks. Genesis AI claims that its hand offers a unique combination of anthropomorphic design, high precision, and AI-driven control that sets it apart.
Another key differentiator is the end-to-end integration of AI model and hardware. By controlling the full stack, Genesis AI can optimize the model specifically for its hand’s capabilities, and vice versa. This vertical integration is reminiscent of how companies like Apple control both hardware and software to deliver a seamless user experience.
Expert Perspectives and Potential Impact
Industry analysts have noted that the ability to perform fine manipulation tasks — such as tying a knot, handling fragile objects, or assembling small components — remains a major challenge for robotics. If Genesis AI’s hand and model can reliably execute these tasks, it could open new markets in precision assembly, laboratory automation, and elderly care.
Dr. Daniela Rus, director of MIT CSAIL and an investor in the company, has previously emphasized the importance of soft robotics and adaptive control. Her involvement suggests that Genesis AI may be incorporating cutting-edge research into compliance and force sensing. Similarly, Vladlen Koltun’s expertise in AI and computer vision could help improve the model’s perception abilities.
The $105 million seed round also reflects a broader trend of increased venture capital flowing into deep tech. While many AI startups focus on software-only solutions, Genesis AI’s hardware-software hybrid approach requires more capital but offers higher barriers to entry. The company intends to use the funds to expand its engineering team, scale data collection efforts, and accelerate development of the full-body robot.
Genesis AI has not announced a specific timeline for commercial deployment, but the release of the GENE-26.5 model and robotic hand suggests that the company is moving from research to productization. If successful, it could become a significant player in the next generation of intelligent robotics.
Source: Silicon UK News