Genesis AI, a startup that raised $105 million in a seed round to build fundamental AI for robotics, announced its first model, GENE-26.5. This comes with a surprise hand. In a demo video, the company showcased a variety of advanced tasks performed by a series of in-house designed robotic hands.
“Models have always been the goal, because better models mean better intelligence,” Genesis co-founder and CEO Zhou Xian told TechCrunch. However, the company quickly realized that it needed control over its hardware. “So we decided to go full stack,” he said.
Other well-funded companies are operating at the intersection of AI and robotics, including Physical Intelligence and Skild AI. Chou also acknowledged that “there are probably 50 or 100 robotic hand companies.” But he and co-founder Théophile Gervais are hoping to get an edge by building it themselves.
The main difference is that rather than the two-finger grippers used by many robotics companies, Genesis’ hands are the same size and shape as a human hand, reducing the gap with real-world situations.
“This allows us to collect far more data and train models that can perform far more tasks than was previously possible,” said Garbett, a former research scientist at Mistral AI and now president of Genesis.
Of all the physical manipulation tasks featured in the video below, Gervet’s personal favorite is cooking. This is because robots have proven they can complete a long series of difficult tasks, such as cracking an egg or slicing a tomato. But Genesis also tasks its robots with tasks such as preparing smoothies, playing the piano, and solving Rubik’s Cubes, a robotics gimmick.
Other tasks, such as laboratory work, are closer to commercial applications of Genesis technology. But what’s happening behind the scenes is just as important. The startup has also developed a sensor-equipped glove that acts as a double for the real-world robot hand and collects data that can be more easily used.
“Our idea was that if we could design a robotic hand that tries to imitate the human hand as closely as possible, we could instantly unlock huge amounts of human data without having to worry about what people call the ’embodiment gap’ in robotics research,” Zhou said.
Others have taken up the challenge. The main novelty is how Genesis pairs this with its models. The current version is named GENE-26.5 for May 2026, but thanks to the simulations developed, Zhou expects there will be many iterations. “The real bottleneck for model iteration speed is evaluation, so this helps speed up model training significantly,” he said.
But beyond simulation, data is the key to training models to help robots perform more tasks. Genesis gloves may come in handy there as well. Jurbet said that unlike cumbersome data collection devices that get in the way, this is as light and easy to wear as the security gloves already used in many industries, and can be produced relatively cheaply.
“We’re currently in negotiations with a number of customers, but a lot of the value of the gloves is that they allow people, whether they’re lab technicians in pharmaceutical companies or manufacturing, to wear data collection devices for the first time when they’re doing their daily jobs,” Gerbe said. This will also be supplemented by “egocentric video data”, i.e. people filming themselves performing tasks.
Still, it remains to be seen whether workers will be willing to train the robots to wear gloves and cameras to replace them, and whether they will receive additional pay for that training. It will be between Genesis customers and its employees, Jurbet suggested. “We haven’t finalized the details yet,” he said.
They may decide not to share that data with the startup anyway, the founders acknowledged. But the startup also has its own path to building a “human skills library” and could pay third-party partners to collect data. The model has already been trained on “a large amount of human-based internet video,” according to the press release, but there is no mention of compensation.
Combining this with a simulation system has the potential to reduce the cost of the technology for real-world applications like the one Genesis has demonstrated. “This is an important milestone for their team and the broader robotics industry,” said former Google CEO Eric Schmidt, who invested in the startup.
In July 2025, just a few months after its founding, the startup emerged from stealth with a $105 million seed round co-led by Eclipse and Khosla Ventures, with individuals like Bpifrance, HSG, and Schmidt, as well as additional backers like Xavier Niel, Daniela Rus, and Vladlen Koltun.
This funding allowed Genesis to increase its workforce. The company has offices in Paris and California, as well as a presence in London. “One of the big reasons we decided to locate in Europe is that there is a very high density of talent across the continent,” says Gerbe. The 60-person team is split between “40-45% in Europe and 50-55% in the US,” and the startup is currently hiring in all three locations.
Aside from hiring, the company will soon unveil its first general-purpose robot, and Chou told TechCrunch that it will be a full-body robot, not just hands. But he insisted the roadmap remains the same.
“Our goal is to build the most capable robotic system,” he said.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
