Robots that learn the task. Hands built to do it.

Our software teaches industrial robots new tasks automatically. It trains each skill through thousands of practice attempts in simulation, with no expert hand-tuning, and grades every result on situations the robot never practiced. It is moving onto real robots now. Alongside it, we build the Dex Hand, a dexterous robotic hand designed to carry out those skills.

See the software The Dex Hand
NVIDIA Inception Program member

01 — What we do

Two products, built together and moving fast.

The software trains the skills. The hardware gives those skills somewhere to go next. The software works with the simple two-finger grippers already on factory robots today, and testing on real robots is its next step.

Phase 1 ready now

Policy training pipeline

Describe a task in plain English and provide a digital model of your robot. The pipeline returns a trained policy, the program that moves the robot through that task, already proven in simulation. No engineer has to hand-tune it.

  • Now: trained policies for individual tasks, proven in simulation, for the standard two-finger grippers most factory robots use
  • Ranked by how often it succeeds in situations it never trained on, never by its own internal score
  • Delivered with its success rate in each test condition, plus video
  • Next: the pipeline itself, licensed to companies that automate their own production lines

We want the pipeline in companies' hands as soon as possible. The step between here and there is proof on real robots. If you run a robot cell on a production line, a pre-order request puts your cell in line to be one of the first.

Pre-order request How it works
Finalizing Phase 2 design

Dex Hand

A dexterous robotic hand with six motor-driven joints, for industrial robot arms and humanoid robots, built to run skills trained by the same pipeline.

  • Precision linear actuators drive each finger, run by an onboard controller
  • Senses joint position and fingertip force, with an onboard IMU tracking the hand’s own motion
  • Ships with a digital model, ROS 2 robot software and trained task skills
  • US-designed and assembled
Specs & roadmap

02 — The opportunity

Trainable skills and dexterous hands turn 4.66 million installed robots into general-purpose machines.

Robot arms already move precisely, and robot vision already sees well. Both are deployed at scale, and what multiplies them is a robot that can be taught a new task quickly and a hand that can carry it out. Today every new part, orientation or production-line change needs an expert to reprogram the robot, and the simple clamp on the end does only the one job it was set up for.

That is why assembly, gathering parts into kits, and loading and unloading machines are still done by hand at most plants. We build both pieces: skills trained without an expert in the loop, and an end-effector dexterous enough to use them.

Dex Hand Phase 2 assembly, open pose

03 — Software

Describe the task. Get a verified policy.

Our pipeline automates the slowest part of teaching a robot a skill. Robots learn new skills through reinforcement learning, practicing by trial and error while a reward formula scores every attempt. Writing that formula usually takes an expert months of tuning, repeated for every task and every part. Following NVIDIA’s Eureka method, our system has an AI language model write candidate rewards, trains each one in simulation, and grades every result by how often it actually succeeds in situations it never practiced.

Every failure is recorded and checked on every future run, so the system gets better with each task. Adding a new task becomes a simulation run, not a hiring plan.

How the software works
Dex Hand mounted on 6-axis arm, cell context
95% Pick-and-place success in simulation, on situations never seen in training
0 Reward settings tuned by hand
~1 day To set up a new two-finger gripper

// Results in NVIDIA Isaac Lab, a physics simulator for robots. Measured with training randomness switched off, at approach distances the policy never trained at. Testing on real robots is the next milestone.

04 — Dex Hand

A dexterous hand, designed alongside the policies that run it.

The Dex Hand is trained by the same pipeline as every gripper we support, so it arrives with skills for the customer’s task already learned, rather than as a bare mechanism the customer must program from scratch.

Not just a digital model and driver software, but trained skills for the target task. The setup our pipeline needed for a commercial gripper is the same setup a multi-finger hand needs, so the work on one carries straight over to the other.

Dex Hand fingertip detail, sensor array
Actuation
6 motor-driven joints, plus 5 linked joints that follow them · precision linear actuators
Sensing
Joint position, fingertip force · onboard IMU for motion and orientation
Policy pipeline
Rewards written by an AI language model, following the Eureka method · trained by reinforcement learning in NVIDIA Isaac Lab simulation · skills carried from simulation onto the real hand
Control
Custom microcontroller built into the hand · self-contained, with nothing extra to mount on the arm
Integration
URDF digital model · ROS 2 packages for standard robot software · trained task skills
Origin
US-designed and assembled · auditable supply chain

05 — Market

The market is already installed, and continues to expand.

4.66M Robots in operation
542K Installed in 2024
700K+ Annual, forecast 2028

Every one of those robots needs its tasks taught. The software is built for that installed base, working with the grippers already in service. The hand reaches the tasks those simple grippers cannot.

// Source: IFR World Robotics 2025 · Market for robot grippers and hands $5.54B (2025) → $10.58B (2030), Mordor Intelligence

06 — Roadmap

Two tracks, moving in parallel.

Software

  • Ready now

    Phase 1 — Trained policies per task

    The pipeline trains and verifies a policy for a given gripper and task, in simulation. Pick-and-place is solved end to end at 95% success on situations held out from training, with no hand tuning; an industry-standard adaptive gripper is going through the same stages now.

  • Critical next step

    Phase 2 — Real-hardware validation

    Real robot cells, with real success rates reported next to the simulated ones.

    This is the step that lets us sell the pipeline. We are pursuing it now, and we need hardware: a gripper, an arm and a real task. If you can offer a robot cell, send a pre-order request.

  • As soon as Phase 2 proves it

    Phase 3 — The pipeline, licensed to companies

    Automation teams run the pipeline themselves on their own robots and tasks. More grippers and hands, more kinds of tasks.

Hardware

  • Complete

    Phase 1 — Mechanism

    Fully 3D-printed and servo-driven. Proved out how the fingers move.

  • Finalizing Design

    Phase 2 — Software test platform

    Aluminum frame, linear actuators, custom controller, full sensor suite. Design being finalized with suppliers and machine shops, built to prove the software against real parts.

  • Next

    Phase 3 — Production

    The hand that ships, designed for volume manufacturing from the start.

We are raising a pre-seed round.It funds both tracks: proving the pipeline on real robots, and putting Phase 2 hands to work on production lines.

Investor inquiry →

Follow along as we build.Results, failures and what they taught the system, on the blog.

Read the blog →

07 — Team

The founders.

Kameron Jones

Kameron Jones

Co-Founder

Lead Engineer — mechanism, tolerancing, manufacturing

Milo Kucia

Milo Kucia

Co-Founder

Software Lead — reinforcement learning, control, firmware

08 — Contact

Start a conversation.

We are raising a pre-seed round to take the pipeline onto real robots and put Phase 2 hands to work on production lines. If you are an investor, a systems integrator, or an automation team with a robot cell that could be our first real-world test, we would like to hear from you.

Pre-order request Investor inquiry