How Do Humanoid Robots Actually Help Humans?
Under every humanoid robot video, somebody asks the same thing. A composite of the versions we keep seeing:
“I’ve watched the backflips, the half-marathon, the robot folding one towel very slowly. Fine. But what does a humanoid actually do for a person? Who is it helping right now, and is it helping me or the company that bought it?”
Most answers skip the second half of that question, and it’s the right half to ask. As of November 2026 the honest answer has three parts. On job sites, humanoids and their cousins already help, mostly by taking weight off human bodies. In homes and in care, the help is promised, partly real, and often has a hidden human attached. And whether a robot helps the worker or replaces her depends less on the robot than on who owns it and what we build around it.
Where They Help Today: Heavy, Repetitive, Hot
The real deployments are unglamorous. Figure’s robots spent an 11-month pilot at a BMW plant and helped build 30,000 cars. Agility’s Digit has moved totes for the logistics firm GXO since 2024. Boston Dynamics’ new Atlas, which lifts 110 pounds, is practicing sequencing car parts at a training center inside Hyundai’s Georgia Metaplant.
The clearest example of help that works isn’t a humanoid at all. Gritt, a San Francisco startup, bolts AI-directed robot arms onto the excavators and pile-drivers a construction crew already owns. With them, an eight-person crew installs 3,000 to 4,000 solar panels a day. By hand, the same eight people place about 800. Gritt’s machines set part of the modules at the 346-megawatt Gibson City solar project in Illinois this summer, and both Gritt and its partner Burns & McDonnell describe the system as one that complements crews.
That is help in the plainest sense: the same people, four times the output, fewer hours with a panel on your shoulder in the sun. The sun part counts. The World Health Organization reports that 2.4 billion workers are exposed to excessive heat, with productivity dropping 2 to 3 percent for every degree above 20°C. A machine that carries the load in that heat spares somebody’s back and lungs.
Where They’re Promised: The Home and the Care Gap
The bigger promise is care, and it rests on a demographic fact. More than 10 percent of Japanese are now 80 or older, and the country projects an 11 million worker shortfall by 2040. China is heading toward more than 20 percent of its population aged 65 and over in the early 2030s. There will not be enough young people to lift, bathe, fetch and clean for everyone who needs it. That is why Asia Pacific leads humanoid adoption: necessity, more than enthusiasm for gadgets.
The machines aren’t ready for that job yet, and the best evidence comes from the companies selling them. In September, Figure tested its Helix 2.5 software in 30 Bay Area homes the robot had never seen, on three chores: tidying a living room, folding towels, making a bed. Trained from scratch, it succeeded on 9 percent of trials. Pretrained on footage from Figure’s Index app, where people in 108 countries are paid to film their own chores, it succeeded on 56 percent. A sixfold gain, and the robot still failed almost every other time.
1X’s Neo, the best-known consumer humanoid at $499 a month, had no independently verified customer delivery as of late October. Weave’s $8,000 Isaac folds laundry “autonomous by default,” but when it gets stuck, a remote human specialist can take over the arms.
From the Podcast: A $30 Robot Cleaning Visit
The November 9, 2026 episode of Minds, Bodies & Terawatts, our daily podcast, put the home promise against a real visit. Its headline: A Reporter Paid $30 for a Humanoid to Clean Her Apartment. A Person in Another Room Was Driving It, the Price Is Low Because You’re the Training Data, and Amazon Is Hiring for a Home Robot.
The episode walks through an MIT Technology Review piece by Grace Huckins, who booked Tau Robotics’ cleaning robot in San Francisco under her real name. The first visit ended early because the remote operator’s equipment overheated. On the second, the robot wiped the counters and cleaned the toilet. Its cleaning wand head fell off three times, and each time it picked the head up and screwed it back on. It also knocked over a closed bottle of olive oil. Tau’s cofounder said vacuuming and mopping already run without a human’s help. Everything else still has a person steering. The same weekend, CNBC reported that Amazon has formed a home robot team; its earlier home robot, Astro, reached only a few thousand customers and never left beta.
Asked whether this takes a house cleaner’s clients yet, the episode says no: the cheapest reliable way to get a clean bathroom is still a human. Then comes the line that answers this post’s title better than any spec sheet: “The line gets crossed one chore at a time, not one job at a time.” The book calls that line the Substitution Threshold, the moment the cheapest reliable provider of a service stops being a person.
Help for the Body, or for the Balance Sheet?
That chore-by-chore crossing is where help and replacement part ways. Chapter 1 of the book, in “What This Means for You,” describes the version where the robot helps: a nurse on double shifts whose robots handle “the physical labor of lifting patients, transporting equipment, and managing supplies, while you do what only humans can: comfort the frightened, interpret the ambiguous symptom, hold the hand of someone who’s dying.” For the construction worker whose back is giving out at forty-five: “Your body stops being the price of admission.”
The robot doesn’t choose that outcome. People do. Agility’s own filing says its robot subscriptions are priced “at a discount to a customer’s fully burdened labor cost,” which means a falling build cost stays with the vendor. Gritt’s roadmap, “solar plants, then everything else,” reads honestly as eight workers becoming six, then four. The same machine that spares the nurse’s back can eliminate the nurse, and which one happens depends on whether survival is still tied to the paycheck. The book’s answer is the Foundation: housing, food, healthcare and energy provided unconditionally, so a robot taking a chore off your hands lands as relief instead of a pink slip.
On unscarcity.ai: The Three Articles Behind This Answer
The Humanoid Robot Revolution tracks who is shipping what, from Unitree’s first quarterly report to Tesla’s late-2027 date for outside customers, and explains why demonstration data, more than hardware, is the real bottleneck. Robots That Build Abundance follows Gritt and the compounding loop that starts when machines build the energy machines. Autonomy Theater covers the remote humans behind “autonomous” robots and argues that this stage is normal, a dimmer moving rather than a scam.
In the Book: The Lightbulb and the Exoskeleton
Chapter 4 of Unscarcity, “Residents and Citizens: Who Gets to Belong?”, has two passages that sharpen what “help” means.
In “The Japanese Model: When Care Became Currency,” Mrs. Tanaka, eighty-two, in an Osaka apartment in 1995, cannot reach the hallway lightbulb because her hips have failed. Her son lives three hours away. A young neighbor changes the bulb, stays for tea, and is paid in a Fureai Kippu, a Caring Relationship Ticket, which she sends to her own mother in Sapporo. The chapter’s observation is that “care is a different currency than cash.” The paid worker watched the clock, the volunteer watched the person, and elderly Japanese preferred the volunteers. A humanoid will change that bulb soon enough. Nothing in its spec sheet covers the tea.
“The Disability Question” runs the other way, toward bodies the machines extend. In the book’s 2052, “Robotic exoskeletons let a paralyzed person pilot construction mechs,” and brain-computer interfaces let a non-verbal person write fusion reactor code. Then the chapter meets Iris, whose locked-in syndrome even those interfaces cannot fully bridge. Her Civic Service is research on her own condition: documenting what works, testing new interface prototypes with the teams building them. In the book, every data point she provides counts as Impact.
More questions like this one are collected on the blog page. Follow the series on Substack.
