French robotics-focused startup Genesis AI has released its first foundational model, GENE-26.5, alongside a human-like robotic hand built entirely in-house. The hand is capable of performing complex tasks such as cooking, playing the piano, and solving a Rubik’s Cube, showcasing significant progress in dexterous manipulation—one of the hardest challenges in robotics.
Company Background and Funding
Genesis AI completed a $105 million (£77.2 million) seed funding round in July 2024, co-led by Eclipse and Khosla Ventures. Additional investors included Bpifrance, HSG, former Google CEO Eric Schmidt, telecoms entrepreneur Xavier Niel, MIT Computer Science and Artificial Intelligence Laboratory director Daniela Rus, and Apple distinguished scientist Vladlen Koltun. The substantial seed round—among the largest for a European robotics startup—reflects growing investor interest in AI-driven hardware.
The company was founded by CEO Zhou Xian and president Theophile Gervet. Gervet previously worked at French AI firm Mistral, which is known for its open-source large language models. This connection highlights the cross-pollination between language AI and robotics, as Genesis leverages transformer-based architectures for robot control.
The GENE-26.5 Model and Robotic Hand
The GENE-26.5 foundational model is designed to power real-world robotic systems. Unlike many robotics models that focus only on software, Genesis chose to build its own hardware to control the full stack more precisely. The result is a robotic hand that closely mimics a human appendage, with five fingers and a flexible palm. Most competing robotic hands use two or three fingers, which limits dexterity and the ability to transfer human demonstration data directly to the robot.
By creating a hand with human-like anatomy, Genesis can collect training data from humans performing tasks—either through video recordings or lightweight sensor gloves—and transfer that data to the robot without complex retargeting. This approach could accelerate learning for tasks like assembly, cooking, or medical procedures.
Sensor Gloves and Data Collection
Genesis has developed lightweight sensor gloves that workers in industries such as pharmaceuticals and manufacturing can wear while performing tasks. These gloves capture fine-grained motion data, which is combined with video footage to train the AI model. The company claims this multi-modal data improves the robot’s ability to generalize to new scenarios.
However, the use of worker data raises ethical questions. Employees might be reluctant to help train robots that could eventually replace them. Genesis has not yet addressed how it plans to manage this tension, but the issue is becoming increasingly relevant as robotics and AI advance in the workplace.
Demonstrations and Capabilities
In a demo video, Genesis showed a pair of its robotic hands preparing a smoothie—picking up fruits, opening a blender, and pouring liquid—as well as playing the piano with coordinated finger movements. The hand’s ability to handle both heavy objects and delicate items suggests a wide range of potential applications, from industrial automation to assistive devices.
Global Presence and Expansion Plans
Genesis AI maintains offices in San Carlos (California), Paris, and London. Its team of about 60 people is split roughly evenly between the US and Europe, with a slightly larger group in California. The company cited Europe’s large talent pool and potential industrial customer base as reasons for keeping a significant European presence.
While the current focus is the hand and model, Genesis is already working on a full-body general-purpose robot. This would integrate the dexterous hand with a mobile base and arm, enabling tasks like cleaning, warehouse picking, or even home assistance. The company’s long-term vision is to create a platform that can be trained on human data and deployed across multiple industries.
Context: The State of Robotic Dexterity
Dexterous manipulation has long been a bottleneck in robotics. While robots excel at repetitive tasks in structured environments (like car assembly), they struggle with the variability of human spaces. Recent advances in AI—especially in imitation learning and reinforcement learning—have enabled robots to learn from human demonstrations. Companies like Tesla (with its Optimus robot), Figure AI, and Boston Dynamics are all pursuing similar goals, but most use grippers or simplified hands.
Genesis’s decision to build a human-like hand is ambitious but could give it an edge in tasks that require fine motor skills. The human hand has 27 degrees of freedom, making it incredibly complex to replicate mechanically and control algorithmically. By solving this, Genesis could unlock new applications in healthcare, logistics, and service robotics.
The seed funding round attracted prominent backers from both the tech and investment spheres. Eric Schmidt, a longtime advocate of AI, and Xavier Niel, a leader in European tech, signal confidence in the startup’s approach. The involvement of Vladlen Koltun, known for his work in computer vision and robotics at Apple, adds technical credibility.
The robotics industry is also watching how Genesis will compete with larger, well-funded players. Figure AI recently raised $675 million, and Tesla’s Optimus is backed by enormous resources. However, Genesis’s focus on a human-like hand and full-stack control might allow it to specialize in niches where dexterity is critical.
Another key factor is Europe’s role in the AI race. While the US and China dominate headlines, Europe has a strong research base in robotics and AI ethics. Genesis’s dual presence in both regions could help it navigate regulatory challenges and access diverse talent.
The company’s next milestones will likely include expanding its team, developing the full-body robot, and proving that its hand can operate reliably in real-world environments. If successful, Genesis could become a significant player in the next wave of robotics—one that brings human-like manipulation to factories, labs, and homes.
Source: Silicon UK News