Genesis AI, a startup that raised a $105 million seed round for developing foundational AI for robotics, has introduced its first model, GENE-26.5, featuring unexpected hands. In a demo video, the company highlighted several advanced tasks executed by the in-house designed robotic hands.
“The model has always been the goal because a better model means better intelligence,” said Genesis co-founder and CEO Zhou Xian. However, the company recognized the need for control over hardware. “So we decided to go full stack,” he added.
Other well-funded companies like Physical Intelligence and Skild AI work at the intersection of AI and robotics. Zhou noted there are “probably 50 or 100 robotic hand companies out there,” but he and co-founder Théophile Gervet hope their unique design will give them an advantage.
A key distinction is that Genesis’ hand matches the size and shape of a human hand, unlike the two-finger grippers commonly used in robotics, narrowing the gap with real-world conditions.
“That lets us collect a lot more data than was previously possible, to train a model that can do many more tasks,” said Gervet, now Genesis’ president. Among the tasks in the video, Gervet’s favorite is cooking, demonstrating the robot’s ability to complete complex tasks like cracking an egg and slicing a tomato. The robots have also been assigned tasks like preparing smoothies, playing the piano, and solving a Rubik’s cube.
Other tasks, such as lab work, indicate potential commercial applications of Genesis’ technology. Behind the scenes, the startup developed a sensor-loaded glove, a real-life double of its robotic hand, for data collection.
“Our idea was that if we could design a robotic hand closely resembling a human hand, we could access a lot of human data immediately without worrying about the ‘embodiment gap’ in robotics research,” said Zhou.
While others have addressed this issue, Genesis’ novelty lies in combining it with its model. Named GENE-26.5 for May 2026, Zhou expects many iterations, aided by their developed simulation. “The real bottleneck for the iteration speed of the model is evaluation. So this helps us speed up model training,” he noted.
Apart from simulation, data is crucial for training models to help robots perform more tasks. Genesis’ glove could aid this process, said Gervet, as it’s light, easy to wear, and inexpensive.
“We’re in talks with customers, and a glove’s value is that, for the first time, it can be worn for daily jobs, whether as a lab technician for pharma or manufacturing,” Gervet said. It would be paired with “egocentric video data,” where people film themselves performing tasks.
It’s uncertain if workers will be pleased to wear such gloves and cameras, potentially training robots to replace them, or if they will receive extra pay for that training. These details lie between Genesis’ customers and employees. “We haven’t nailed the details yet,” Gervet acknowledged.
Regardless, customers may choose not to share data with the startup, but Genesis can build its “human skill library” through third-party data collection. The model is already trained on “massive amounts of human-based internet videos.”
Combined with its simulation system, this could reduce costs for real-world applications like the one demonstrated. “This marks an important milestone for their team and the robotics industry more broadly,” said Google’s former CEO, Eric Schmidt, an investor in the startup.
In July 2025, shortly after its formation, Genesis emerged from stealth with funding co-led by Eclipse and Khosla Ventures, backed by Bpifrance, HSG, Schmidt, Xavier Niel, Daniela Rus, and Vladlen Koltun.
The funding allowed Genesis to expand its team, with offices in Paris and California and new operations in London. “A significant reason for being in Europe is the dense talent across the continent,” Gervet explained. The 60-person team is split with “40-45% in Europe and 50-55% in the U.S.,” and the startup is hiring in all locations.
The company plans to unveil its first general-purpose robot soon, described by Zhou as a full-body robot, not just hands. However, the roadmap remains the same.
“Our goal is to build the most capable robotic system,” he stated.