Enact raises $30M pre-seed in Zurich to give robots force and touch, backed by Yann LeCun

Enact, a Zurich robotics startup, said on Oct 22 that it has raised a $30 million pre-seed round as it comes out of stealth. The company announced the round in a post on its LinkedIn page; it did not name a lead investor or disclose a valuation.

The backers it named are the venture firms redalpine, Notion Capital and firstminute capital, together with Yann LeCun, Arthur Mensch of Mistral, Andrew Lonsberry of Path Robotics and Thomas Wolf of Hugging Face.

What Enact says it is building

Enact describes its product as an artificial nervous system for robots. Instead of adding tactile skins or force-torque sensors, its software reads force and touch from the signals of the actuators a robot already has, so the robot can sense contact in real time. The company argues that vision and language models are becoming commodities and that robots need this sense of physical interaction to learn thousands of fine manipulation tasks efficiently, aiming first at work where the value lies in handling variation rather than repeating one high-volume task. It has not named customers, robot partners or a product date.

How the round compares

In MEKAWIRE's database Enact's round is the fourth-largest seed-stage robotics financing of $15 million or more in the past 12 months, behind Walden Robotics, Dream Labs and Generative Bionics. It is also twelve times the $2.5 million pre-seed that fellow Zurich startup mimic raised in May 2024 for dexterous robot hands. Method: amounts as disclosed by each company; Enact's $30 million divided by mimic's $2.5 million.

The bet against extra sensors

The approach runs opposite to Zurich neighbour [Bota Systems](/companies/bota-systems), which sells dedicated multi-axis force-torque sensors, and to the tactile-skin makers covered in our [robot touch explainer](/news/robot-touch-explained-tactile-sensors-dexterous-hands-who-leads-sharpa-rhoda-jodell-wonik-aidin-tesollo-20261016). Sensing through actuator signals avoids fragile hardware, but it depends on how much contact information motor currents and joint readings carry, which Enact has not yet shown publicly. The next signals are its first published results and the robots it chooses to run on. Company page: [Enact](/companies/enact-intelligence).