EN
Back to the archive

The archive · Space, Robots, Defence · Strategic decision · 2025–2026

Enigma's robot-interface bet: $71M seed, 100+ robots open to the public

Israeli lab bets robot adoption's bottleneck is the interface, not model smarts; opens 100+ robots to anyone online after a $71M seed.

Enigma

The betRobot adoption is blocked by the interface, not model smarts — so open 100+ robots to the public and let real interactions define both UX and training.Building

What the business is

An AI-for-robotics research lab building robot-control foundation models and interfaces, tested live by anyone online via 100+ robots in hangars in Israel and California.

Starting capital$71M seed led by Index Ventures and Ribbit Capital, with Sarah Guo's Conviction Partners participating

How it started

Co-founders Jonathan Jacobi and Gal Niv met as teenagers at hacking competitions and served together in Israel's Unit 8200, where they did cybersecurity research. Last year they set out to apply their engineering background to robotics, a field they had no direct experience in, betting it held tech's most exciting unsolved problems. They recruited what Jacobi calls their 'smartest friends': alumni of top AI labs, math Olympiad winners, and people convinced to drop out of PhD programs.

What happened

Enigma raised a $71M seed led by Index Ventures and Ribbit Capital, with Conviction's Sarah Guo participating. On 2026-07-27 it emerged from stealth and opened more than 100 of its proprietary robots — arms and models both built from scratch — to anyone online. The robots draw with paintbrushes, spar with swords, and mix chemistry flasks in hangars in Israel and California; the public experiment tests whether people prefer text, audio, video demonstrations, or tap-and-drag control.

No ending yet — it is still running.

Background

Enigma emerged from stealth on 2026-07-27 betting that what keeps robots out of daily life is not how smart they are but how people tell them what to do. Founders Jonathan Jacobi and Gal Niv — Unit 8200 veterans who met at teenage hacking competitions — raised a $71M seed led by Index Ventures and Ribbit Capital and immediately opened more than 100 of their own robots to anyone online.

Rather than only training models to master tasks, the lab studies how untrained humans naturally direct machines: text, voice, a video example, or drag-and-drop. Its robots draw with paintbrushes, spar with swords, and mix chemistry flasks in hangars in Israel and California; the company says the arms and the underlying models were built entirely from scratch. Jacobi's target metaphor is the car volume knob — control so intuitive nobody reads a manual.

Index partner Shardul Shah frames the outsiders' edge: robotics insiders start from dexterity and teleoperation, while Enigma starts from 'what's the ultimate experience?'. No valuation, customers, or revenue were announced with the round, though the startup says it is already partnering with companies in healthcare, logistics, and entertainment.

What has to be true

  • Interface-first flips the industry question from 'can the robot do it' to 'will anyone use it' — a different and cheaper bottleneck to attack.
  • Opening 100+ robots to the public makes every online interaction a model-training sample and a UX experiment at the same time.
  • The founders deliberately chose a field they were not insiders in, arguing outsiders ask questions incumbents skip.
  • A $71M seed with no disclosed valuation or revenue shows investors priced the thesis and team, not traction.

What can be applied

When an industry fixates on capability, the binding constraint can be the user interface. Opening the product to the public turns every interaction into both UX research and training data.

Aftermath

As of 2026-09-02 Enigma is running its public experiment at robots.online with 100+ robots and says it is partnering with companies in healthcare, logistics, and entertainment without disclosing use cases. It has not published revenue, customers, or a valuation, and the business model remains open-ended: the interface it discovers is meant to define both the product and the next training phase of its foundation model.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases