Figure.02's Groundbreaking Breakthrough

Dear Readers,

Some breakthroughs can only be recognized in hindsight – others unfold their full impact in real time. Helix from Figure clearly belongs to the second category. In just four seconds, the humanoid robot picks up deformed packages, smooths out plastic film for barcode scanners, and learns directly from the real world of work – not from the laboratory. What is happening here is not just engineering. It is a preview of the next industrial revolution.

Why is this so significant? Because Helix shows how close we really are to merging AI software and the physical world. Thanks to sensors, memory, and adaptive behavior, the result is not just a better robot – but a learning colleague who no longer just performs tasks, but understands them. The number that says it all: With just 60 hours of demonstration data, Helix increases its barcode recognition rate from 88% to 95%.

And Figure is not alone: Unitree is accelerating training with Roboverse by a factor of 30, and Agility is even questioning the need for five fingers. The debate about the meaning and form of humanity in machines has only just begun.

Perhaps Robot Tuesday will be more than just a column in the future – perhaps it will become the day of the week when we renegotiate the future of work.

In Today’s Issue:

• Figure's Helix update is setting new standards for humanoid robots in logistics

• One Robot for Every Human?!

• AI advancements alongside critical ethical concerns

• The revolutionary tool that lets you endlessly reshape your video footage

• And more AI goodness…

All the best,

Figure.02's groundbreaking breakthrough

Figure has unveiled Helix, an update that sets new standards in robotics: The humanoid robot can now handle complex tasks in real logistics environments, such as gripping soft plastic packaging, deformed packages, and flat envelopes. And it does so in just 4.05 seconds per object – around 20% faster than before. This is made possible by a combination of visual recognition, a historical state memory, and tactile feedback via a sensor system.

What makes it special is that Helix does not learn in the lab, but directly in the real working world. With the help of demonstration data – around 60 hours is enough – the system trains itself to grasp, adjust, and correct objects in the same way humans do. The robot acts increasingly autonomously: for example, it recognizes when a film is obstructing scanning and smoothes it out on its own.

This is highly relevant for the AI community: Helix combines modular software with physical intelligence and demonstrates how scalable data and adaptive control are ushering in a new era of humanoid assistance systems.

Why it matters: Helix shows that robots are already capable of reliably performing complex, tactile tasks. This is a breakthrough for humanoid systems – and a sign that intelligent machines are reaching practical maturity.

Source:

🔗 https://www.figure.ai/news/scaling-helix-logistics

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

2. Master AI tools, tutorials, and news in just 3 minutes a day

3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

Sign up and start learning AI

“As we scaled from 10 to 60 hours of training data we saw the time/package drop from 6.3 to 4.3 seconds and barcode scanning accuracy improve from 88% to ~95%” — Figure

Work for 60 minutes straight!

In a new video from robotics company figure, the Figure.02 model works on an assembly line for 60 minutes without a break. It impressively demonstrates how far we have already come. The robots are comin

Unitree x Roboverse

Unitree Robotics will use Roboverse as its central platform for developing intelligent robots. This infrastructure combines simulation, datasets, and benchmarks, accelerating the training of robotic behaviors by 30 times. The goal is to develop humanoid robots that are more versatile and powerful. At the same time, an open developer community with hackathons and global collaboration will be established. A demo already shows impressive progress.

Not all fingers necessary?

”Agility CEO Peggy Johnson says humans have five fingers, but you rarely need all five. Robots need dexterity and fine usability, but not necessarily five fingers.