Enigma raised $71 million in seed funding to build more intuitive ways for humans to control robots, marking one of the largest early-stage rounds in robotics this year.

The Israeli startup, co-founded by former Microsoft engineer Jonathan Jacobi and cybersecurity researcher Gal Niv, takes a different approach to robotic AI than competitors like Physical Intelligence and Figure AI. Rather than focusing purely on model capabilities, Enigma studies human-robot interactions to develop better control interfaces.

Index Ventures and Ribbit Capital co-led the round, with participation from Sarah Guo's Conviction Partners. The funding will support Enigma's research into foundation models that can execute tasks they were never explicitly trained to handle.

Testing human-robot interaction at scale

Enigma is launching a large-scale experiment allowing anyone worldwide to interact online with over 100 proprietary AI robots housed in facilities across Israel and California. The robots can draw pictures with paintbrushes, engage in sword fighting, and perform basic chemistry experiments by manipulating flasks and liquids.

The company developed both the robotic arms and their underlying AI models entirely in-house, according to the founders. This contrasts with approaches used by other robotics companies that study web videos, run computer simulations, or collect motion data from humans wearing sensor-equipped gloves.

Jacobi, Microsoft's youngest-ever employee who was recruited by Wiz founder Assaf Rappaport, met co-founder Niv during teenage hacking competitions. Both later served in Israel's elite Unit 8200 cybersecurity division before deciding to tackle robotics despite lacking direct experience in the field.

The team includes alumni from top AI labs, mathematics Olympiad winners, and several researchers who left PhD programs to join the startup. Enigma emerged from stealth less than one year after its founding.

The company plans to use the funding to expand its robot fleet and accelerate research into what it calls "a different kind of robotic brain" based on human interaction patterns rather than traditional training methodologies.