Enact Intelligence

Verified

Track: stealthWebsite ↗𝕏 @enact_intel

potential 76 · scoop 68

Enact Intelligence builds actuators with embedded high-frequency force sensing for robot dexterity. Former 1X controls lead and Physical Intelligence engineer Manuel Galliker raised a $30M pre-seed in Zurich.

Scores run 0 to 100. Potential: how likely the firm is to become a real winner, judged across the six dimension scores and held consistent with them, whether or not anyone has heard of it. Scoop: how unlikely a well-read tech investor is to have heard of it yet. High means it has not reached the outlets that reader follows, whatever the firm's quality or age. How our top potential scores did.

ShareXLinkedIn
Manuel Galliker led controls on 1X’s NEO humanoid and co-authored Physical Intelligence’s π0.5 VLA model before raising $30M in one of Europe’s largest pre-seeds from Zurich.

Snapshot

What they build
Builds robots that sense force and touch directly through their motor actuators (augmenting models with proprietary actuator data), rather than relying primarily on cameras or separate sensors.
Who buys
Industrial partners in manufacturing, infrastructure, retail, and food processing.
Comparables
1X · Physical Intelligence
Sub-industry
physical AI / robotics actuators and sensing

Ex-1X (led software for NEO humanoid) and Physical Intelligence alum; team includes engineers from DeepMind, Amazon Robotics, and Helsing.

How we found them

Surfaced in issue
Oct 26, 2026
Announced
Oct 22, 2026 (stealth reveal)
Founded
2026
Round
pre seed · $30,000,000
Investors
Redalpine, Notion Capital, firstminute Capital, Yann LeCun, Arthur Mensch
HQ
Zurich, Switzerland
Discovery track
stealth

Originating post ↗

Dimension scores

  • Team: 85 out of 100
  • Product: 70 out of 100
  • Traction: 45 out of 100
  • Market: 80 out of 100
  • Moat: 60 out of 100
  • Capital: 90 out of 100

Analysis

Galliker built a 12-person team in Zurich. Multiple engineers came from 1X, Physical Intelligence, and ETH Zurich. Later hires arrived from DeepMind, Amazon Robotics, and Helsing. Philip Kneis, formerly at lead investor redalpine, joined after the round to run growth and operations.

The hardware places force sensors inside force-transparent motors that sample above 1 kHz. Models process force as a primary input alongside vision through multi-scale architectures. The stack targets high-mix European manufacturing lines rather than volume humanoid production.

redalpine led the $30M round. Angels include Yann LeCun, Arthur Mensch, and Thomas Wolf. No public benchmarks or customer deployments exist yet. Execution risk stays high.

What they do

Enact Intelligence builds robots that sense force and touch in real time directly through custom motor actuators, feeding proprietary proprioceptive data into AI models, rather than relying primarily on cameras or add-on sensors, targeting contact-rich dexterity tasks such as cable insertion, seal fitting, and screw starting in variable industrial settings.

Why now

Foundation models (vision/language/VLA) have matured sufficiently (e.g., post-π0 work) to serve as strong priors that can be adapted rather than trained from scratch for common sense; the dexterity/contact bottleneck has become the recognized limiter after initial humanoid locomotion demos; European manufacturers face pressure for flexible, high-mix automation where vision-only approaches struggle; actuator/sensor integration advances and teleop data pipelines now make closed-loop force feedback practical at startup scale. The $30M pre-seed timing aligns with a broader physical AI funding wave.

Founding story

Galliker founded Enact in March 2026 in Zurich shortly after his Physical Intelligence stint, drawing directly from frustrations encountered while leading controls on 1X’s NEO humanoid and working on generalist VLA models: vision provides position but fails at the moment of contact (e.g., grip force on soft or misaligned objects), limiting dexterity in real-world tasks. He explicitly chose Zurich over the Bay Area for its denser hardware/robotics engineering talent pool versus Silicon Valley’s software/AI strength, aiming to combine imported AI experience with local hardware capabilities.

Competitive position

Obvious incumbents: 1X (prior employer, vision + emerging touch hands), Figure AI, Boston Dynamics, Unitree (primarily camera/position-based control). Closest startup comparables: Physical Intelligence (models Galliker helped build), Path Robotics (angel investor, welding focus), Genesis AI (hand-level manipulation). Enact occupies a distinct wedge, integrated hardware (custom actuators) + models + data modality, positioned for any form factor but initially targeting industrial assembly variance rather than general humanoids or pure software models.

Defensibility

Emerging moat from proprietary actuator designs that embed high-precision force sensing without extra hardware, combined with integrated model architectures and data pipelines that treat force as a first-class modality; talent concentration from 1X/PI/DeepMind/ETH provides execution edge; potential switching costs once deployed in customer lines. Early-stage reality means moat is largely prospective and could be challenged by well-resourced incumbents replicating the sensing approach.

Network density

Of the initial ~12-person Zurich team, multiple engineers hail from overlapping prior organizations (1X, Physical Intelligence, ETH Zurich), with additional hires from DeepMind, Amazon Robotics, and Helsing. Philip Kneis (former deeptech investor at lead VC redalpine who worked on the Enact deal) joined post-funding to lead growth/operations, creating a direct investor-operator bridge. This suggests at least 3, 5 early members with shared 1X/PI/ETH pedigrees in a small founding cohort.

Early team signals

Of the initial ~12-person Zurich team, multiple engineers hail from overlapping prior organizations (1X, Physical Intelligence, ETH Zurich), with additional hires from DeepMind, Amazon Robotics, and Helsing. Philip Kneis (former deeptech investor at lead VC redalpine who worked on the Enact deal) joined post-funding to lead growth/operations.

Traction

  • $30M pre-seed closed (one of Europe’s largest)
  • Live website
  • Consistent multi-outlet coverage (SiliconANGLE, TNW, Sifted, EU-Startups, NZZ, etc.)
  • 12-person team with elite hires + active recruiting (AI researchers, operators)
  • Philip Kneis onboard for growth
  • Plans announced for 5× team growth by end-2027 and first industrial deployments (manufacturing, infrastructure, retail, food processing)

Opportunities

  • Continued progress in multimodal foundation models enabling efficient adaptation of video priors with force data
  • Scaling of teleoperation demonstration pipelines for rapid task learning (potentially shortening training from months to days)
  • European policy/funding tailwinds for advanced manufacturing automation and onshoring
  • Potential first-mover data advantage in force-rich industrial tasks
  • Broader physical AI investment momentum providing follow-on capital

Risks / red flags

  • No public performance benchmarks, test results, or demos have been released despite strong claims on efficiency/reliability gains
  • Full-stack hardware (actuators) + AI development at pre-seed with a 12-person team carries significant execution and capital-intensity risk
  • Single named founder creates key-person dependency
  • “In talks” with industrial partners but zero confirmed customer logos, deployments, or revenue signals
  • Extremely large pre-seed for a 7-month-old company may reflect hype-driven valuation or aggressive burn
  • Regulatory exposure in industrial robotics (safety standards) not visibly addressed yet

Hidden bets

  • Bespoke “force-transparent” motors sampling at high frequency (>1,000 Hz)
  • Multi-frequency/multi-scale model architectures that adapt existing video/foundation models
  • Full data pipelines from teleoperation demonstrations
  • “Artificial nervous system + cortex” framing
  • Hiring AI researchers focused on pre-training from scratch (harder than fine-tuning)
  • Remote teleop operators in Zurich for demonstration data collection
  • Strategic bet on Zurich as hardware talent hub + European high-variety manufacturing wedge

Key relationships

  • Yann LeCun (AMI Labs) – angel investor
  • Arthur Mensch (Mistral) – angel investor
  • Thomas Wolf (Hugging Face) – angel investor
  • Andrew Lonsberry (Path Robotics founder) – angel investor
  • Marc Pollefeys (ETH Zürich) – angel investor
  • Thomas Hofmann (ETH Zürich) – angel investor
  • Paolo Cerruti (Northvolt) – angel investor
  • Elias Kleimann (Wingtra) – angel investor
  • Philip Kneis (redalpine) – post-funding growth/operations lead
  • Harald Nieder (redalpine GP) – lead investor

Founders

Manuel Yves Galliker, CEO & Founder
Education: BSc Mechanical Engineering (Robotics, Systems & Control), ETH Zurich (2014–2018), MSc Mechanical Engineering (Robotics, Systems & Control), ETH Zurich (2019–2022), Visiting researcher, Caltech AMBER Lab (Aaron Ames)
Key projects: π0.5: a Vision-Language-Action Model with Open-World Generalization (arXiv:2504.16054 / CoRL 2025 preprint), Real-time execution of action chunking flow policies (K. Black, M. Galliker, S. Levine; NeurIPS 2025), Bipedal Locomotion with Nonlinear Model Predictive Control: Online Gait Generation using Whole-Body Dynamics (Humanoids 2022, best paper finalist), Whole-body Humanoid MPC work, Master’s thesis on data-driven dynamics modeling (2021)
Prior: 1X Technologies / Research Engineer Robotic Learning / Jun 2024–Jan 2025; 1X Technologies / Team Lead Controls & Embedded / Sep 2023–Jun 2024; 1X Technologies / Senior Robotic Controls Engineer / Sep 2022–Sep 2023; Physical Intelligence / Member of Technical Staff / Mar–Aug 2025
𝕏 @manumerousGitHub

Citations