By Christopher Mims | Photography and Video by Tony Luong for WSJ
The Steve Jobs of robot vacuum cleaners is ready to unveil his next chapter.
It's an expressive, furry, doe-eyed creature with huge paws and a dog-like demeanor. He hopes you'll want to take it home, snuggle with it and cherish it as an essential member of your household.
Colin Angle led iRobot for nearly three decades, during which the company invented the Roomba, and with it the entire category of home robo-vacs. He left in 2024. The first product of his new startup, Familiar Machines & Magic, has something no Roomba ever had: four legs.
It isn't just a robot pet, he insists. It's not a smart assistant either. It won't talk, but it will react to your actions and feelings in an emotionally intelligent way, he says.
"I want to build a relationship between human and machine that is fundamentally different than the relationship between the Roomba and a customer," says Angle. "I want to feel like the machine I'm building actually cares about me."
Angle and his team call it the Familiar -- the same term witches use for their supernatural animal companions. It's meant to be first in a new category of emotionally intelligent robots, trained to respond appropriately to our tone of voice, body language and overall vibe. This connection would enable robots to act in support roles that require a high level of trust, he says.
In the short term, Familiar Machines plans to market its products to people who want to monitor their loved ones, such as an aging parent or other dependent adult living alone. The company is also keen to sell Familiars to people who'd like to support their own well-being. Further along, the company hopes to license its emotionally intelligent AI to other companies, says Angle.
The Familiar unveiled Monday at The Wall Street Journal's Future of Everything conference is still an early prototype, so Angle isn't disclosing its price or availability, or even how much the final product will resemble its current look when it hits the market.
We're not far from a future where we have packages delivered by autonomous drones and we work alongside robots, including humanoid ones. We might also share the monitoring of loved ones with chatty robot companions.
Imagine if your Apple Watch grew legs and could follow you around, not merely vibrating to remind you to stand up every half-hour, but playfully nuzzling your hand to remind you to do so. The upsides are obvious, though so are some concerns: We're seeing overreliance on technology to address our emotional needs lead to a litany of harms.
When pressed on this, Angle and partners emphasize that the Familiar's wordless interactions inherently soften its approach, making it more like R2-D2 than M3GAN.
Another robot pet?
While it's early days for Familiar Machines, the company's goals are ambitious. The history of companion robots is littered with e-waste carcasses, including Sony's Aibo, Ugobe's sauropod Pleo and the souped-up smart speaker known as Jibo. The biggest reason robot pets don't succeed is that flesh-and-blood pets are plentiful and do a great job addressing their humans' needs.
This time is different, insists Chris Jones, a company co-founder and iRobot's former chief technology officer.
To explain why, he cites pet waste -- not only a downside of real pet ownership, but something early Roombas were infamous for transforming into a homeowner's worst nightmare.
Just a few years ago, the AI on the Roomba struggled to recognize objects on the floor like socks or a pet's whoopsie. Today's vision-based and generative-AI systems, by contrast, can take in entire scenes in ways unimaginable when Jones was working on robo-vacs.
"It took a huge amount of work to recognize just a very small number of categories of objects that Roomba might encounter," says Jones. "Today, with the latest AI models, we essentially get an enormously rich understanding of the world around us." That includes not just objects but people: facial expressions, gaze and pose, gestures and the broader context of their behavior.
Jones describes two scenarios for when a Familiar's owner comes home: In the first, the person bends down with arms wide open, ready for a hug. In the second, the person is rushed, with an armload of groceries, no time for robo-snuggles. The Familiar will know whether to run up or to hang back.
And unlike our smart speakers or other stationary robots and sensors, the Familiar will be able to amble about our homes, watching us all the time. This will help it build what its makers call a "social graph": how people in a household relate to one another, and what their relationship is to the Familiar itself.
The privacy question
A robot pet that can walk into any room of your house, always regarding you with cute camera eyes and sensitive microphone ears, could easily threaten our privacy.
The company's founders insist that they're taking a privacy-first approach. By default, it doesn't send any data to the cloud, says Ira Renfrew, Familiar Machines' co-founder and chief product officer. As with the Roomba, it will ask permission if it ever needs to access the internet, he adds.
Processing all of that AI inside of a miniature-schnauzer-size body is no small feat. But the latest silicon makes the task easier than before. Edge AI (so called because processing is done at the "edge" instead of in the cloud) is something we already see in smartphones, like when Siri processes your verbal requests using the iPhone's onboard silicon. Since it doesn't speak, the Familiar doesn't have to run a gigantic large language model. Instead, it's using smaller models specialized to its tasks.
Because a Familiar is designed to observe and support, its emotional processing is focused on recognizing someone's mood, based on facial expressions, body language and tone of voice. It's also supposed to recognize people it knows and to be careful about approaching people it doesn't.
Is a person happy or sad? Are they having an argument? When does a sad person want comfort...or solitude? The Familiar must be programmed to handle these scenarios, like an autonomous vehicle is programmed to handle tricky intersections.
All of a Familiar's behaviors are intended to be approachable and inoffensive, says Morgan Pope, a Familiar Machines roboticist who spent almost eight years at Disney Research. Some of the ways the robot moves are based on the moves of dogs and other animals: how close it comes to a person, the speed at which it approaches, even the way it bats its eyes and twitches its ears.
Training a robot on the intricate timing of initiating interaction with a new person is an "enormously hard" problem, says Pope.
Emotional labor
We now have countless examples of how wrong things can go when people lean too heavily on AI for emotional support. And yet Angle is explicit about his goals with Familiar Machines: He wants users to bond with it.
Whether or not this company succeeds, it raises a question that robotics companies will be grappling with for years: How do you responsibly build a system that knows us intimately, into which we place ever greater trust?
When I press Renfrew about this, he turns philosophical. "People will, and should, know that they're interacting with a machine," he says. "I think that most of us feel warmth and connection to objects in our lives, right? I'm particularly fond of these pens. I'm particularly fond of my Ioniq 5."
Familiar Machines is betting that an emotionally intelligent robot will tap in to our tendency to grow attached to machines, and that this will make its creations better helpers.
In summing up his aspirations, Angle mentions the film "Jurassic Park." To him, the white-bearded CEO who revived the dinosaurs is unfairly characterized.
"To me, he's the hero of that movie, not the villain, because he was trying to take something that was always smoke and mirrors and turn it into something magical and real," he says. "That's some of the sentiment behind what we're doing."
Hopefully, the Familiar Machines story has a different ending.
Write to Christopher Mims at christopher.mims@wsj.com
(END) Dow Jones Newswires
May 04, 2026 12:00 ET (16:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments