Enigma raises $70M seed to make robots easier to control

A new robotics lab arguing that the real bottleneck in manipulation robots is not the model but the interface just picked up an unusually large seed round. Enigma, per TechCrunch, came out of stealth with $70 million led by Index Ventures and Ribbit Capital, with Sarah Guo's Conviction Partners also in. That is a lot of money for a seed, and it signals what the investors think the moment is: models are now good enough to move a robot arm around, and the harder problem is telling one what to do without a PhD in robotics.

Co-founder Jonathan Jacobi, described in the reporting as Microsoft's youngest-ever employee, framed the thesis with a dishwashing analogy: if getting a robot to put things away takes fifteen minutes of instruction, most people give up and do it themselves. He and co-founder Gal Niv met on hacking competitions as teenagers and later served together in Israel's Unit 8200. The stated goal is to make controlling a robot feel more like adjusting a car's volume knob.

To collect data on how ordinary people actually try to direct machines, Enigma is opening more than 100 of its own robots, housed in Israel and California, to a public online experiment where anyone can operate them to paint, sword-fight or run chemistry experiments. It has also signed unnamed partners in healthcare, logistics and entertainment. Index partner Shardul Shah's pitch is that the founders' lack of a traditional robotics background is the point, and that it leaves 'more room for originality.'

A $70M seed at this stage buys a bet, not a validated product, and it lands in a busy funding tape where we have logged 387 AI rounds in the last 90 days. The reporting does not describe what the interface actually looks like, whether it wraps existing vision-language-action models or is trained from scratch, or what those enterprise partners have actually committed to beyond a logo slide. A public sword-fighting demo is a great funnel for interaction data, and it is also an efficient way to generate viral failure clips.

If the thesis holds up, the beneficiaries are the buyers who cannot hire a robotics team: hospitals, warehouses, small manufacturers who need a fleet operator, not an engineer. That is a bigger market than the humanoid-form-factor race everyone else is chasing, and the interesting thing to watch is whether the interface layer proves defensible on its own or gets absorbed by the model labs upstream.