Enact Intelligence exits stealth with €26.68 million pre-seed funding

Zurich-based physical AI startup Enact Intelligence raised €26.68 million ($30 million) in pre-seed funding on October 22, 2026, and exited stealth, eu-startups.com reported. The round included redalpine, Notion Capital and firstminute capital. Individual investors included Arthur Mensch, Yann LeCun and Andrew Lonsberry.

According to the announcement, Enact aims to build robots that use force and tactile data in addition to vision. This represents an approach that changes the data design to address grasping and contact tasks that are difficult to handle with camera-image-centered perception alone. The reported participants include redalpine, Notion Capital and firstminute capital, as well as Yann LeCun and Arthur Mensch, pointing to capital flowing toward connecting robotics foundation models with real-world data.

What is new in Enact Intelligence’s move is not just the funding, but its decision to exit stealth by foregrounding a concept that supplements vision-centered robot perception with force and tactile data. Integrating perception and control remains a challenge in the existing robotics industry, but the company says it will rethink the full stack starting with “high-precision force data from existing actuators.” This suggests it is seeking to differentiate itself through the design of hardware inputs, not software alone. A related article from us, Dream Labs launches robot control API, covered the higher-level layer of controlling robots through natural language. Enact, by contrast, emphasizes a lower-level data foundation focused on sensory input at the end effector rather than higher-level APIs. Together, the two show that the competitive front in robot intelligence is expanding from the “command layer” to the “contact-understanding layer.” However, the announcement does not make clear which types of robots, tasks or environments the company will target first. Structurally, Enact’s strength lies in using force data for training, but its effectiveness will depend on the quantity and quality of that data, how it is collected and whether training can be reproduced on real machines. On the capital front, the €26.68 million raise is large for a pre-seed round and could give the company time for research and development and hiring, but it has not yet reached the revenue or mass-production stage. Key questions going forward include what sensor configurations it will use to collect data, whether it can be retrofitted to existing robots and how it will integrate with vision-centered models.

In Japan, many areas involving industrial robots and manufacturing equipment require precise contact and force control. If designs based on nonvisual data, such as Enact’s, become more widespread, the method of collecting input data could become a competitive factor alongside the software layer for companies working on existing robot controls and sensor design.