We already delegate many household tasks to robots, cooking, cleaning, air conditioning or security. But each of them is dedicated to their own plot of the home. Wouldn’t it be better to have just one, but let him take care of everything? That’s what California-based robotics company Sunday asked. And the answer is Memo (yes, the name may not be very lucky in Spanish, but at its root is memory, not simplicity).
This little robot with wheels, precise arms and a very ambitious mission has just emerged from the laboratory to enter our daily lives with the aim of taking care of some of the most routine household tasks: load the dishwasher, fold clothes, organize and even prepare a espresso. But the most interesting thing is how he learns to do all this.
Where does “learning” come from? The key is a peripheral: the Skill Capture Glove, a glove designed by Sunday to record how people perform household chores. According to the company, they have used it in more than 500 real homes to collect data on how people move, clean, organize and manipulate objects.
With this strategy they have rcollected around 10 million episodes of real household routines. This immensity of data (not virtual, not synthesized) is precisely at the core of Memo’s design. According to Tony Zhao (co-founder and CEO), many home robots fail because they have been trained in overly controlled environments, not in real homes, with all the unpredictability that entails.
Thanks to this, Memo can adapt to the messy and changing nature of a home and perform what they call “long horizon” tasks: sequences that require several steps, a sense of context and the ability to adapt. For example, cleaning the table, picking up dishes and loading them into the dishwasher, folding clothes, especially light items like socks, making coffee…
As for the design, Sunday didn’t want a tall, bipedal robot. Memo stands on a base with wheels, which gives it greater stability and reduces the risk of falls, even if it loses power. It has a telescopic structure that allows you to vary its height (for example, reach a countertop or look under tables).Its body is covered in soft silicone, which makes it more friendly and “domestic”: it does not look like a cold machine, but rather an assistant with a safe and gentle presence in the home.
Why can this way of training robots change everything? The database it uses, real homes, is a novelty and if we add to that the interaction in those homes, with humans and their routineswe can add flexibility to adapt to tasks and schedules, without the need for programming.
Even so, Memo is not a miracle machine, without limit or problem: the tasks it does are very specific: it will not clean large spills, nor replace a person for complex tasks. Although the glove includes many routines, does not guarantee that all homes are the same: the 500 homes used for training They can be very different from others, and despite this, there is a large margin of error when Memo arrives at new homes.
There is the question of privacy: If the robot learns from your movements and how you distribute your objects, how does it manage that data? Sunday says he keeps this in mind, but it’s something anyone inviting one into their home should keep in mind.
Regarding maintenance, Memo is going to use sensors, motors, wheels… all of that wears out, and bringing a robot home also means thinking about its long-term maintenance. The company is recruiting 50 homes for a foundational beta, which will begin in late 2026, which means we’ll soon see Memo in real homes, not just at demos or tech fairs.