I've always believed that some of the best ideas come from questioning the conventional way of doing things. When we were building Roomba, the idea of having a robot autonomously clean your home seemed pretty unconventional. Today, millions of people live with one. Now, at @FamiliarMxM, we're exploring another possibility: what if robots could do more than perform tasks? What if they could understand and respond to people, develop relationships over time, and become a meaningful presence in our lives? It's a very different way of thinking about what robots are for. I talked about this approach to problem-solving at TEDxFargo: youtube.com/watch?v=obDJOg__…

Meta’s Muse and OpenAI’s Dots show AI starting to move beyond answering questions and into taking action. Put that kind of agency into a machine that shares our physical world, and the challenges and risks increase further. A robot in a home, hospital, school, or workplace has to read people. It has to understand tone, attention, mood, relationships, timing, and personal boundaries. It has to know when to approach or stay back, and how to behave differently around a child, an older adult, or a group of people. A lot of that judgment requires social and emotional intelligence. There’s enormous investment going into teaching machines how to perceive, reason, move, and manipulate the physical world. Yet the industry is still underestimating something fundamental: machines that operate around people need to understand people. That’s the AI platform we’re building at Familiar Machines & Magic: social and emotional intelligence for #PhysicalAI. We believe it will be critical for the category to succeed.

Data centers may be leaving earth. Every major technological era eventually creates pressure on infrastructure that was built for the previous one. Electrification changed grids. Cars changed cities. The internet changed telecom infrastructure. Now, AI is changing compute and energy.  With #Suncatcher, @Google is exploring putting AI compute into orbit, using solar-powered satellites - next week, we'll see the first in-orbit test of AI chips. There’s a lot of invention still to come around the systems that make AI possible. Our own AI team at @FamiliarMxM has been rethinking how to maximize performance for #PhysicalAI running on the edge. I’m excited to see what other assumptions get challenged as we figure out what it takes to bring Physical AI into the world at scale. Image credit: NASA

Part of what differentiates @FamiliarMxM is how we approach our work. We obsess over the experience. We move quickly. We spend where it buys progress. And we believe innovation only counts when it ships.