Enact Intelligence exits stealth with €26.68M to build robots that feel, not just see

Enact Intelligence (Zürich) has raised €26.68 million in a Pre-Seed round that draws three institutional funds — redalpine, Notion Capital, and firstminute capital — plus a set of angels whose names carry specific technical weight: Arthur Mensch, co-founder of Mistral; Yann LeCun, founder of AMI Labs and former Chief AI Scientist at Meta; and Andrew Lonsberry, founder of Path Robotics. The company was founded in March 2026 by Manuel Yves Galliker, who led software development for the humanoid robot NEO at 1X before moving to Physical Intelligence, and exits stealth with a team of 12 and industrial pilots already in progress.

The proprioceptive bet

The dominant paradigm in physical AI is vision-centric: robots learn from video data what manipulation looks like, then attempt to replicate it. Enact's founding premise is that this approach fails systematically at the tasks factory operators actually need — gripping irregular objects, applying calibrated pressure in assembly, navigating environments where visual cues are ambiguous or occluded. Its answer is a proprioceptive architecture: capturing force data directly from the actuators inside the robot, building a distinct data modality that vision-only models cannot replicate without the same hardware.

The 12-person founding bench draws from DeepMind, 1X, Amazon Robotics, Helsing, Physical Intelligence, and ETH Zürich. Philip Kneis, formerly a DeepTech investor at redalpine, has joined as growth lead — an investor-to-operator move that says something about conviction.

"Dexterity is the next frontier in robotics, and Enact is tackling it with a new data modality," said Dr. Harald Nieder, General Partner at redalpine. "Starting with high precision force data from bespoke actuators, they are rethinking the entire stack from the hardware up — leveraging orthogonal data to improve existing models and build a data moat."

Why LeCun and Mensch

The angels on Enact's cap table are not generalist cheque writers. LeCun has argued publicly that sensorimotor data — the kind Enact is building its architecture around — is the missing ingredient for general physical AI. Mensch co-founded Mistral on the thesis that differentiating on architecture and data quality beats brute-scale data collection; the bet on Enact applies the same logic to robotics. Both have track records of spotting the missing data modality before the consensus does, which makes their co-investment a signal rather than a coincidence.

The €26.68 million will fund a fivefold expansion of the team by end of 2027, alongside hardware development and rollouts with industrial partners in manufacturing, retail and food processing. The company exits stealth with named industrial partners and a measurable product claim; what the capital needs to produce is a first cohort of deployments where force-sensing robots demonstrably outperform vision-only alternatives on factory floors that care about throughput, not research benchmarks.