Model precision test. Trained with 2 minutes examples.

Preparing a photo for the website

A few complex bin-picking grasps.

Benefits of using a reliable VLA model on example of our customer’s task. Instead of building special tooling, we use the model to separate the top magnet from the stack directly in the grippers. In the demo, the model moves the stack up, one magnet at a time, visually finding the right height. It also corrects the slight randomness in the magnets’ positions so the top magnet ends up in the same place, where it’s easy to pick it off. When the stack is almost empty, the model sees it as the right gripper covers the stack, and later it will go get a new batch, but for now, in this test, it just throws it away. No coding, except saving points 1x speed, autonomous

I didn't expect our grippers to evolve into cute little hands. But it’s the most functional design for now.

Strong magnets can make the environment self-resetting.

Customer self-onboarding into our system. They are deploying the robot themselves, consulting with us in chat. They will collect data (the least time-consuming part) and train the Loop Model using our tutorial.

A Salesforce Tower full of MicroFactories could replace Shenzhen for the US. A single Microfactory occupies about 3×2×2 feet (12 cubic feet). The Salesforce Tower in San Francisco is about 22 million cubic feet, so it could fit around 1.83 million MicroFactories. Meanwhile, Shenzhen has 1.8 million people employed in manufacturing. Sure, factories need extra space for elevators, material storage, etc., but Shenzhen also ships only 20% of its output to the US, and not everyone is doing physical work, many hold managerial positions.

Preparing the second MicroFactory unit so they can work in a line. Real-world Factorio is coming.

Let’s use MicroFactory as a ScienceFactory. Standardized replicable cells, industrial grade precise arms, interchangeable grippers for tool use

Robotic insight 3 There’s a widely held view among leading robotics thinkers: the current priority is to automate some commercially profitable work to get real-world data. If you set the goal as “the first profitable deployment that also generates data” you get a few insights: - People with industry experience may be better positioned to nail this than academic researchers. - You can try to find non obvious initial domain, that’s easier to deploy in practice, while still provide useful data.

Robotic insight 2: One undocumented feature of VLA models is that they can generalize pretty well from very little data, if you run them on good hardware. Here, with 35 demonstrations (5 mins in total), robot can pick up small pcb from any position and at any rotation angle.

The advantage of arms with industrial internals is that they don’t wobble, so the AI model can control it faster by just multiplying frames per second. Here’s we multiplied the FPS by 3× compared to teleoperation (180 instead of 60).

It’s quite underestimated that with just 8–10 hours of training, you can have a robot that performs commercially useful tasks already today. Spend a little time once, and you get a mechanical worker that reliably does the job 24/7 for years.

Industrial UR5e arm inside our box for size comparison. Industrial arms are good because they are rigid and have low backlash, so they do not shake during operation. However, they are quite bulky. One of the reasons to build own arms is to combine the best of both worlds: compact size and stability.