Enigma, a research lab less than a year old, emerged from stealth with a $70M seed round led by Index Ventures and Ribbit Capital, with participation from Sarah Guo of Conviction Partners. The startup is taking a fundamentally different approach to the hardest problems in robotics AI: rather than building foundation models purely through web videos, simulations, or motion-capture gloves, Enigma wants to study how humans naturally engage with robots, hoping that those interactions will lead to intuitive interfaces and possibly a new kind of robotic brain.

Enigma is launching a large-scale experiment that lets anyone in the world interact online with more than 100 of its proprietary AI robots. The robots, housed in hangars in Israel and California, can paint with a brush, fight each other with swords, and perform simple chemistry experiments such as picking up and mixing flasks of liquid. Enigma claims it built both the robotic arms and their underlying models entirely from scratch.

The founders are Jonathan Jacobi (Microsoft’s youngest-ever employee, recruited by Wiz founder Asaf Rappaport) and his longtime friend Gal Niv. The two met as young teens in hacking competitions and later served together in Israel’s Unit 8200 doing cybersecurity research. They had no direct robotics experience but believed the field held the most exciting unsolved problems in tech. Their team includes alumni from top AI labs, math Olympiad winners, and several people who dropped out of PhD programs.

Jacobi told TechCrunch that Enigma‘s goal is to make human-robot interaction completely effortless, as intuitive as adjusting a car’s volume knob. The startup’s public test will evaluate different communication methods: text, audio, video examples, or tap/drag/drop interfaces. Jacobi admits the experiment is very open-ended. The company hopes to discover not only superior interfaces but also better ways to train its foundational AI model. While Enigma already partners with companies in healthcare, logistics, and entertainment, its business use case remains unclear.