August 26, 2026
New Atlas reported this week on a San Francisco startup taking pre-orders for a humanoid household robot, the Nori A3, priced at $1,688. That number is the headline, and it should be. Every humanoid robot demo of the last few years has come wrapped in a price tag aimed at industrial buyers or well-funded research labs, usually five or six figures. A sub-$1,700 price point, even for a machine that reportedly leans on tele-operation rather than full autonomy for a lot of its tasks, is the first real signal that humanoid hardware is drifting toward consumer reach.
Here's the part worth sitting with, though: tele-operation matters. If a chunk of what this robot does is a remote human operator driving it through your laundry pile, you're not really buying autonomy yet -- you're buying a very early, very cute interface to labor that's still mostly human. That's not a knock on the product; it's a reasonable way to ship something real while the perception and manipulation stack catches up. But businesses evaluating any 'humanoid robot' pitch right now should ask the boring question nobody wants to ask at a demo: how much of this is the robot, and how much is a person in another room? The gap between marketing and reality in robotics has always lived in that answer, and it's the same gap that separates a genuine automation strategy from a very expensive toy. If you're building internal workflows around automation claims, it's worth applying the same scrutiny you'd bring to any workflow automation vendor promising more autonomy than the tech currently delivers.
IEEE Spectrum's piece on Ollobot and the broader AI companion robot category makes a point that's easy to miss amid the hardware hype: the first wave of companion robots, going back to around 2017, had personality and voice interaction but were narrow enough that novelty wore off fast. Owners talked to them for a week, then let them gather dust. The pitch this time is that better AI closes that gap -- more natural conversation, more persistent memory, more reason to keep engaging past the honeymoon period.
I'm skeptical that better conversation alone solves the retention problem, and I think that skepticism is the right starting point for anyone evaluating this category, whether for elder care, customer-facing use, or just curiosity. Novelty decay isn't a software bug -- it's what happens to almost any device that doesn't earn a spot in someone's daily routine through actual utility. A humanoid or tabletop companion that only talks is competing with a phone that already does everything else in your life. The companion robots that stick will probably be the ones that fold into a genuinely useful task -- reminders, monitoring, light physical help -- rather than ones selling connection as the entire product. That's a lesson that applies well beyond robotics, honestly: any tool marketed on emotional engagement rather than a clear job to be done tends to face the same drop-off curve, which is part of why we've written before about companion bots struggling to hold attention once the novelty fades.
Put these two stories together and you get a useful, unglamorous takeaway: the humanoid robotics industry is finally being forced to answer for cost and for genuine utility, not just capability demos. That's healthy. A $1,688 price point invites real households and small businesses to kick the tires in a way that a $150,000 research platform never did, and that scrutiny will surface tele-operation dependencies, battery limits, and reliability problems faster than any lab benchmark. Robohub's recent piece asking how green robots actually are points at a related blind spot -- as this hardware gets cheap enough to proliferate, questions about manufacturing footprint and end-of-life waste stop being theoretical and start being procurement questions. None of that means humanoid robots aren't coming for real household and light-commercial work. It means the honest timeline is longer and messier than the demo videos suggest, and the businesses that win will be the ones asking hard questions about autonomy, durability, and total cost before they buy in.
If a $1,688 humanoid robot showed up at your door tomorrow, what's the one chore you'd actually trust it to do unsupervised -- and what would it take to earn your trust on the rest?
Sources