More than two decades after helping define consumer robotics with the Roomba, Colin Angle is returning with a different kind of ambition. At The Wall Street Journal’s Future of Everything conference, he introduced a new company, Familiar Machines & Magic, and outlined a vision that moves beyond task-oriented machines toward something closer to artificial life.
The company’s core idea is a category it calls “Familiars,” physically embodied AI systems designed not just to perform functions but to build ongoing relationships with people. These systems are meant to perceive their environment, develop consistent behavioral patterns, and adapt over time through repeated interaction. The emphasis is less on what the machine can do in isolation and more on how it behaves in the presence of a person.
This reflects a broader shift in how the robotics industry is being framed. Much of the current investment in physical AI is focused on industrial capability, from warehouse automation to humanoid labor systems. The assumption is that progress is defined by dexterity, autonomy, and efficiency. Angle’s argument is that this represents only part of the opportunity. The other half depends on interaction, where machines are not just operating in human environments but directly engaging with human behavior.
That shift introduces a different set of design constraints. Systems intended for interaction need to communicate intent, maintain continuity, and respond in ways that feel legible over time. According to Familiar Machines & Magic, this is where current approaches fall short. Screen-based interfaces and chat-driven AI can simulate conversation, but they lack physical presence, which changes how people respond to them. The company is betting that embodiment matters, particularly in contexts involving emotional support, companionship, or everyday guidance.
The first Familiar introduced at the event reflects that design philosophy. It is a quadruped system built specifically for interaction rather than labor. Its structure allows for expressive movement across 23 degrees of freedom, supported by a touch-sensitive outer layer, a vision system, and an integrated audio setup. The system runs on an onboard multimodal model designed to process visual, auditory, and contextual inputs in real time, allowing it to respond with behavior rather than just output.
What distinguishes the approach is not the hardware itself but the emphasis on continuity. The system is intended to remember interactions, adapt to individuals, and present a consistent presence over time. This positions it less as a tool and more as an ongoing participant in a shared environment. The design avoids humanoid mimicry and instead focuses on approachability and expressiveness, suggesting that familiarity does not require replication of human form.
The team behind the company brings experience from the earlier wave of consumer robotics. At iRobot, they helped deploy more than fifty million devices into homes worldwide, turning a niche category into a standard household technology. The new venture builds on that foundation but shifts the goal from utility to interaction. It also draws on talent from organizations like Disney Research, MIT, and Boston Dynamics, combining expertise in robotics, AI, and human-machine interaction.
Familiar Machines & Magic is positioning itself around what it calls consumer physical AI, a category defined less by what machines can accomplish and more by how they are experienced in daily life. The underlying claim is that the next phase of robotics will not be won by capability alone, but by systems that people are willing to live with over time.
Join the Discussion