When Alexis Cholas lost his arm while working as a volunteer combat medic in the Hospitallers Medical Battalion in Eastern Ukraine, he thought his “life was over”.
But six months later, equipped with a sleek black AI-assisted robotic arm made by Ukrainian-US startup Esper Bionics, he was able to return to the frontlines. The device meant the typically life-changing injury, in fact, wasn’t, he tells Sifted.
The prosthetic limb, dubbed the Esper Hand, is built with specially designed muscle sensors that the company says allows users better control and dexterity than others on the market.
But the startup’s not in the prosthetic limb business, says founder and CEO Dima Gazda. Esper is using its robotic arm as a testing platform to develop next generation sensors for a “mass market” — which could be used on wearable devices or implanted inside the body, he tells Sifted.
Gazda is banking on that approach unlocking “way bigger revenue” than the company could if it was focused on only manufacturing prosthetics. Esper has just raised $5m in a funding round led by YZR Capital, with participation from EBRD and U.Ventures.
AI-assisted robotic arms
Esper has been developing its robotic arm since it was founded in Kyiv in 2019, and deployed its prototype to its first user the following year. The startup moved its headquarters to New York in May 2022 — which was delayed a couple of months by Russia’s invasion of Ukraine — and also has offices in Berlin. Forty of its 55-strong global team, including the bulk of the company’s R&D, is still based in Kyiv.
Most robotic prosthetic arms use two sensors to control all functions of the hand — detecting movement from the flexor and extensor muscle groups, says Gazda. “It’s like controlling a multifunctional device with only two buttons.” Esper’s device, however, is controlled with the equivalent of around “20 buttons”.
While Gazda is unable to share any specific details about the sensors as he’s worried others may copy them, he says they’ve been developed in-house and collect more data than others on the market, which can then be used to better train the machine learning algorithms the company’s also developed that support how the device moves.
Esper has also built a cloud-based software platform that complements its robotic arm, collecting and processing data and housing its AI control algorithms — which, along with the sensors and prosthetic arm, Gazda calls the “data collection architecture”.
The long game
Esper currently sells its bionic arms via insurers in the US, who pay $22k apiece. That’s about 10% more expensive than the average robotic prosthetic. It sells the device for $7k in Ukraine — which is also the cost of production, meaning the company doesn’t make a profit on those units — funded by donations and the government.
But making revenue from the sale of its bionic arms is “not the main focus”, says Gazda.
“If you are trying to make revenue from day one, you will build products that exactly fit requirements to be reimbursed by insurance companies — but you won’t develop new health technology,” Gazda tells Sifted.
Instead of attempting to take market share with its prosthetic arm, Esper’s activity so far has been an exercise in collecting data that can be used to develop next generation sensors for wearables devices and implants in the future.
These advanced sensors are bulky things when they’re first developed — and wouldn’t fit into the discreet wearables on the market today. But to develop designs and functionality, and eventually reduce their size, they need to be tested on a whole lot of users.
For a wearable like the Oura ring to collect as many data points as Gazda thinks is possible, for example, it would need to be the size of an armband right now — and thousands of people aren’t going to wear that, he says.
But there is a sizeable group of people who are happy to wear a large wearable device — people with prosthetics, Gazda tells Sifted. “It’s smarter to help people who already have big wearable devices like prosthetics, and by doing this you have the expertise to build sensors for smaller devices [in the future].”
The next steps
Of the 120 prosthetics currently in use, 80% are in Ukraine and the remainder are in the US; Esper wants to sell 300 more of its robotic arms on the US market in the next 12 months, and a big portion of the funding will be spent on upscaling its manufacturing capacity.
But Gazda thinks that most of the startup’s future revenue will come from licensing the sensors it develops out to other wearable manufacturers. It hopes to go to market with that offering in 18-24 months.
While Esper is just developing muscle sensors on its robotic arm at the moment, he says that working with other manufacturers — and using the data collection architecture the company is building — it could develop sensors that pick up a multitude of data points in the future.
These could include things like heart rate, blood oxygen saturation, barometric pressure and temperature, Gazda tells Sifted — along with others he says he can’t reveal at the moment.
