Figure AI just dropped Figure 03, and holy moly: this might be the first humanoid robot actually built for your living room.
Think of it as the Model T of robots. The big difference from previous versions? This bad boy was designed from the ground up for mass production. No more hand-built prototypes that cost a gazillion dollars each.
Here’s the deal: Figure 03 can fold laundry, load your dishwasher, clear tables, and even water plants. It’s covered in soft, washable fabric (think knit sweater vibes) instead of cold hard metal, making it 9% lighter and way less likely to accidentally clothesline you in the hallway.
What makes it tick? Figure’s proprietary AI system called Helix. It’s basically a vision-language-action system that let it learn new tasks by watching humans. The robot learned to fold towels from just 80 hours of video footage (almost as much time as your Nephew spends learning TikTok dances).
The tech upgrades from Figure 02 to 03 are actually wild
Cameras with 2x frame rate, 60% wider field of view, and 75% less latency.
Palm cameras in each hand for close-up viewing (like having eyes in your hands).
Tactile fingertip sensors that detect forces as light as 3 grams (the weight of a paperclip).
Wireless charging through its feet; just steps onto a charging pad.
Better speakers and microphones for voice control.
Why this matters
Figure raised $1 billion at a $39 billion valuation from investors including Nvidia, Jeff Bezos, OpenAI, and Microsoft. They’re building a new factory called BotQ that’ll pump out 12,000 robots per year initially, targeting 100,000 over four years.
Oh, and TIME just named it one of the Best Inventions of 2025.
The reality check: As flashy as this demo is, this robot isn’t actually ready for home use just yet. CEO Brett Adcock admits they’re “not there yet” and hopes to nail it by 2026. After all, the robot was only finished a week before the demo. And during TIME’s visit, the robot kept dropping laundry and couldn’t pick it back up.
This is something we wondered, too: if the robot is carrying a box, and then accidentally drops it, could it react quickly enough to catch it mid-air? That’s the demo we wanna see, Brett!
The debate: Many roboticists argue humanoid isn’t the right form factor. Hexapod robots (think six-legged spiders) might be more versatile for certain tasks, though they’re probably further out for consumer use; that said, here’s a pretty epic one under construction right now:
Here’s the ultimate bet
If AI scaling laws apply to robotics the same way they did to ChatGPT, we might be closer than we think. Every robot uploads terabytes of data for continuous learning. More data = smarter robots = your dishes finally getting loaded correctly… and maybe catching that falling plate faster than you can?
Editor’s note: This content originally ran in today’s newsletter send from our sister publication, The Neuron. To read more from The Neuron, sign up for its newsletter here.
The post Is Figure 03 the ‘Model T’ moment for robots? appeared first on eWEEK.
No Responses