The Roomba Creator Is Building a Robot That Wants to Be Your Friend

Colin Angle co-founded iRobot, the company that made the Roomba vacuum cleaner famous. He ran the company for 30 years before leaving after Amazon canceled a planned acquisition. Now he is back with something completely different: a new startup called Familiar Machines & Magic that is building home robots designed to be companions and friends rather than just cleaning machines.

The company showed its first prototype at a Wall Street Journal conference. The robot is about the size of a dog, has movable eyebrows and ears, and can express emotions through its face. It moves around your house on its own and uses artificial intelligence to have conversations with people, learning their preferences and developing its own personality over time.

From Cleaning Robots to Companion Robots

Angle's first company, iRobot, made robots that did one job: clean your floors. They did not interact much with humans. The Roomba was simple and reliable.

His new company is taking a different direction. The Familiar robot is designed to perceive people, adapt to them, and interact in ways that feel natural. It looks like a blend of a bear, an owl, and a golden retriever. The robot runs its AI brain on its own hardware, not by sending information to computers in the cloud. This choice suggests the company is concerned about privacy — a real issue with many smart home devices that collect data and send it over the internet.

Angle brought together a team of people he worked with at iRobot, including former CTO Chris Jones and Ira Renfrew, who previously worked on Amazon's delivery robot project. The team also includes software engineer Colin Soguero and creative director Morgan Pope, according to AP News.

The company says the robot could be useful for companionship, entertainment, eldercare, helping people around the home, and supporting parents with young children, according to reports from The Verge.

The startup is well-funded and is looking to raise $30 million, TechCrunch reported. That much money suggests the company will need to spend heavily on building and perfecting the hardware before it can sell robots to customers.

Why Now?

Over three decades of covering technology, I have watched successful entrepreneurs take lessons from one era and apply them to the next. Angle did it before: he figured out how to make robotic home automation work during the early 2000s, when computing power and sensors were improving fast. Now he is applying those instincts to a new world where AI is becoming more powerful and people are more comfortable with smart devices in their homes.

Companion robots have been attempted before. Sony made a robot dog called AIBO, but it was very expensive and did not catch on. Boston Dynamics has built impressive robots, but they focus on commercial uses, not consumer homes. Startups like Anki and Jibo tried the companion robot idea and struggled to find enough customers.

The timing may be better now. Consumer comfort with AI has grown — people use Alexa, ChatGPT, and other AI tools regularly. Homes are more connected than ever. Battery technology is better. The costs of manufacturing robots are dropping. These changes together might finally make consumer companion robots practical and affordable enough for real homes.

What Could Go Wrong

Building companion robots is harder than building cleaning robots. A Roomba succeeds if it cleans your floor. A companion robot has to do something much trickier: it has to make you feel like talking to it is worthwhile, and it has to feel consistent and genuine across many conversations.

The hardware is complicated and expensive to build. Consumer willingness to pay premium prices for a robot that keeps you company is not yet proven at scale. Previous attempts have shown that what people want from a robot in theory is different from what they are willing to pay for in practice.

Privacy is another concern. A companion robot that lives in your home and listens to conversations would need strong safeguards around how it collects and uses data. The company's choice to run AI processing on the device itself, rather than in the cloud, is a good sign. But the details of how the company will protect users' privacy remain unclear.

The broader market for consumer robots is still unproven territory. Amazon invested billions in robotics and still shut down several projects, including the delivery robot that team member Renfrew worked on. That history shows both the opportunity in this space and the real risk of failure.

The real test for this company will not be whether the team can engineer a good robot — Angle's track record suggests they can. The harder question is whether they can design software and AI behavior that actually makes people want to interact with it day after day, and whether the business model for ongoing AI services can work financially.

What This Means

If Familiar Machines & Magic succeeds, it could show that home robots have finally become practical enough for ordinary people to own and use. That would be a milestone in technology — we have been hearing promises about helpful home robots for decades, and they have not arrived yet.

For the robotics industry more broadly, Angle's decision to move from robots that perform tasks to robots that provide companionship could influence where other companies invest their money and effort. It signals that the hard technical problems of building robots that move reliably and perceive their surroundings may finally be solved well enough to focus on the harder problem: designing interactions that feel meaningful to humans.

The company is based in Woburn, Massachusetts, part of a long tradition of robotics innovation in the Boston area. That region has produced some of the world's leading robotics companies and continues to have the engineering talent and supply chain companies need.

As the prototype moves toward a real product, we will learn a lot about whether people actually want AI companion robots in their homes. The answer will help shape what happens next in robotics and AI.