A humanoid robot named Memo just picked up a screwdriver, a glass, and a ladle—three objects it had never seen before—and handled each one with the kind of casual dexterity you'd expect from a person, not a machine.
This matters because most robots are brittle. Hard-code them to pick up a coffee mug, and they'll pick up that exact mug, every time, in the exact same way. Show them a slightly different mug and they freeze. Sunday Robotics, a California-based firm, took a different approach: instead of programming Memo with thousands of specific instructions, they built it to learn the way humans do—by watching and imitating.
How Memo Actually Learns
The robot comes equipped with a Skill Capture Glove. A human puts on the glove and demonstrates a task: folding socks, wiping a table, making espresso. Memo watches, learns the movement patterns, and stores them in its AI. The company calls this its growing "Skill Library." Unlike traditional robots that need a programmer to intervene every time the environment changes slightly, Memo adapts. It's designed to work in messy, unpredictable spaces—homes, hospitals, retail environments—where flexibility matters more than perfect repetition.
We're a new kind of news feed.
Regular news is designed to drain you. We're a non-profit built to restore you. Every story we publish is scored for impact, progress, and hope.
Start Your News DetoxThe robot itself is only built from the torso up, mounted on a wheeled platform, with two hands doing all the work. It's topped with a silicone shell and no sharp corners, designed to be safe around people. The company even offers customizable hats in different colors, which feels almost quaint until you remember we're talking about a machine that can now handle objects it's never encountered before.
Memo has come a long way in a year. Last year it had one hand and could arrange shoes. Now it handles dishes, does laundry, and makes coffee. The progression suggests the learning curve is accelerating.
The Bigger Picture
This doesn't mean imitation learning will replace all programming. Factories and warehouses will keep using rigid, highly optimized robots—they need perfect consistency and safety guarantees. But for the unpredictable spaces where humans actually live and work, imitation learning is filling a real gap. It's easier to train, faster to deploy, and more flexible when things don't go exactly as planned.
Sunday Robotics plans to launch a beta program in late 2026, inviting select families to try Memo for free. That's when we'll find out if learning from watching humans actually translates to reliable help with real household chaos.
