Humanoids That Help People, Not Just Impress Them
Published:
Originally published on Substack.
I don’t want a humanoid that wins a demo.
I want one that quietly makes someone’s day easier — the kind that shows up, does its job, and makes life a little better without needing applause.
That’s why I keep saying: don’t get obsessed with dexterity alone. But if the team has the ability to make it, I am happy with that.
Human-level hands are a moonshot, and chasing that headline too early distracts from the real mission — delivering consistent, meaningful value in the real world.
If a robot can move totes safely for eight hours, restock shelves without drama, or carry a box from receiving to line-side on time, that’s already a milestone worth celebrating. Every reliable task completed is a step toward a more capable, more human-centered future.
Rodney Brooks has been a much-needed voice of reason in this field. He’s right about two things: true human dexterity is far tougher than it looks in a deck, and shipping robots that work every single day is a completely different game from building cool prototypes.
But here’s where I see things a little differently: the deciding factor isn’t whether a founder “comes from robotics.” Some of the best teams today come from both worlds — AI and robotics — and that’s exactly what makes them powerful. What truly matters is whether the team has the humility and curiosity to learn from one another — AI experts listening to robotics experts , robotics experts listening to AI experts — and whether they’re willing to build from the ground up: mechanics, controls, safety, reliability, together!
It’s that shared willingness to learn, iterate, and refine — in labs, factories, and real deployments — that separates hype from progress. Background helps, yes. But humility, teamwork, and respect for fundamentals matter far more.
What Really Matters
It’s not founder pedigree; it’s team behavior. The best teams listen. They take advice from those who’ve shipped robots, who understand actuators, transmissions, thermal limits, and field conditions. They let real-world data shape their priorities. Whether they come from AI, robotics, or somewhere completely different, that mindset wins every time.
Dexterity is a program, not a product. Keep pushing manipulation research, but don’t let it hold back progress. Let robots master the stable, repetitive, high-impact tasks first. Every cycle in production, every safe handoff, every hour of uptime is a building block toward greater dexterity tomorrow.
General intelligence won’t override physics — but it will accelerate progress. Foundation models can help robots see, plan, and reason far better than before. They can make adaptation and learning faster. But they still need great hardware — joints, power systems, structure — to make those smarts matter. The teams that respect both AI and physics will be the ones to move humanity forward.
I’m more than optimistic about humanoids — I believe deeply in their potential. When done right, they won’t just change industries; they’ll quietly transform daily life.
The first real wins won’t look like sci-fi spectacles.
They’ll look like fewer back injuries in warehouses. Calmer night shifts in hospitals. Safer factories and cleaner disaster zones. And more people spending time on work that actually matters.
That’s the impact I care about — robots not replacing people, but lifting the weight off their shoulders.
If the leading teams keep their egos in check, stay close to the ground, and build from fundamentals — from actuators to controls to autonomy to UX — we’ll get there faster than most expect.
Humanoids don’t have to be perfect to be profound.
They just need to show up, do useful work safely, learn from their mistakes, and keep improving.
And that’s exactly the kind of steady, human-centered progress I’ll always bet on.