Enigma raises $71M on the argument that the robot problem is an interface problem

Enigma, a robotics research lab less than a year old, emerged from stealth on July 27, 2026 with a $71 million seed round — unusually large for the stage. Index Ventures and Ribbit Capital led, with Sarah Guo of Conviction Partners participating.

The founders are Jonathan Jacobi and Gal Niv, who met at teenage hacking competitions and served together in Israel's Unit 8200. Neither is a roboticist.

If you had to do your dishes and spent 15 minutes explaining to a robot where to put everything, everyone reaches the point of 'Forget it, I'll just do it myself.' Right now, everyone is at that point — even with the most capable models. — Jonathan Jacobi, co-founder, Enigma

The thesis and the experiment

Rather than scaling capability through web video, simulation or sensor-glove motion capture, Enigma is studying how humans interact with robots in order to derive the right interface — and, it argues, possibly a different architecture for the robot's brain.

To generate that data it is launching a public experiment letting anyone worldwide teleoperate more than 100 of its proprietary AI robots online, housed in hangars in Israel and California. Demonstrated tasks include painting with a brush, sword-fighting each other and running chemistry experiments mixing liquid flasks. The company says it built both the arms and the models from the ground up.

Index partner Shardul Shah put the bet plainly: "Jonathan and Gal are outsiders — they're not roboticists. It affords them more room for originality." Enigma is already working with companies in healthcare, logistics and entertainment.

Why it matters

The field's dominant assumption is that capability is the bottleneck and interfaces will follow. Enigma is betting the reverse — that a robot capable enough to be useful is already here, and what is missing is any tolerable way to tell it what you want. Jacobi's dishes example is the whole argument in one sentence.

Opening a fleet to public teleoperation is also the most efficient data-collection scheme anyone has proposed. It costs the operators nothing, it produces exactly the interaction data the thesis requires, and it surfaces failure modes no internal test would generate. The obvious risk is that internet-scale operators are not representative of paying customers.