Based on the Agility Robotics blog post: “The Realistic Pathway to Home”, May 26, 2026.
When one of the industry’s biggest players openly says that home humanoids are not coming anytime soon — and explains why — it’s worth reading carefully. Agility Robotics, the company behind the industrial humanoid Digit, has published a rare piece of plain talk: a sober breakdown of what separates a viral YouTube clip from a real household robot. There are three barriers — capability, cost, and safety. And none of them can be cleared with stage promises.
The vision is familiar to everyone. A robot that folds your laundry, helps with the yard, cleans up after your toddler, and brews coffee on the side. Social-media demos make it feel like science fiction is about to move in. The reality, Agility argues, is far more complicated.
“Don’t let a teleoperated robot fool you”
The first barrier is what the machines can actually do. And here Agility points to a trap that is easy to fall into: the boom in large language models has created the illusion that humanoids are about to follow the same trajectory. They are not. Language models were built on a vast, ready-made resource — essentially the entire written output of humanity, freely available online. For a robot moving through physical space, no such resource exists.
For a humanoid to act on its own in the real world — what the industry calls “embodied AI” — it needs a completely different kind of data: the forces involved in movement, varying lighting, joint limits, speeds, safety boundaries, contact dynamics with surfaces. All of this has to be gathered from scratch, in simulations and real-world deployments, with robots trained through demonstration and reinforcement learning.
And here Agility gives the reader a warning that carries real weight, coming from a manufacturer itself:
“Don’t let the teleoperated humanoid fool you. You might see a robot operating in a public space crowded with people, giving you the perception that humanoids can already autonomously perform these complex behaviors — but it’s likely controlled by a person somewhere out of sight. This isn’t a fully autonomous robot that could scale in a helpful way. It’s a deceptively disguised trick.”
“But Digit already works at Schaeffler and Toyota,” you might object. It does. A factory, though, is a controlled environment: point A to point B, regulated conditions, a closed work cell. The home, Agility writes, is one of the most chaotic places a robot can possibly operate — conditions change day to day, even minute to minute, and the range of tasks is far broader than on a warehouse floor.
To convey just how big that gap is, Agility draws a comparison with self-driving cars. A car moves forward and back, left and right, and has to stay on the road. A humanoid has to go wherever a person can go and do whatever a person does. Even self-driving developers are still collecting data with massive fleets of vehicles — sometimes with a human at the wheel, sometimes with a remote operator standing by to intervene. And for humanoids, the number of scenarios is orders of magnitude greater.
The Price Has to Be Comparable to a Family Car
The second barrier is money. For a household robot to become a mass-market product, its price has to be roughly that of an ordinary family car. Today, it costs many times more.
Agility’s chosen approach is openly pragmatic: bring costs down first by tackling simple, safe tasks in logistics and manufacturing — moving totes between workstations, palletizing and depalletizing. These sectors, the company estimates, have millions of unfilled jobs, and the work there is “high impact and low complexity” compared with anything in a home. The more units shipped to industrial customers, the cheaper each new robot becomes — and the closer the economics get to a point where moving into the home is realistic. Skipping industry and heading straight for the home, Agility writes plainly, is not a realistic goal.
“95% Reliability Is Not Enough”
The third barrier — and, in Agility’s own view, the most important one — is safety. Here the company is blunt: a household robot that works “95% of the time” is not fit for a home. The remaining 5% is a margin of error that is acceptable on paper but unacceptable in daily life. The benchmark is the reliability of the home appliances that have long since faded into the background — the microwave, the washing machine, the robotic vacuum. We no longer see them as complex engineering, precisely because they just work.
And then comes the most sobering passage. Right now, Agility admits, the robots are still being refined. That means accidents. That means robots fall. Inside a factory, behind safety barriers, that is acceptable. Inside a home, it is not. And the question is not a philosophical one about a robot’s intent — it’s about a “statistical calculation of risk of human injury, validated by third parties.” In practice, that means formal Hazard and Risk Assessments (HARAs) and a mitigation plan for every risk: how to avoid pinch points when moving a turned-off robot, how to handle errors of judgment when the machine decides whether to lift something heavy or hot, how the robot tells a person from an object — and how it calculates the appropriate force to apply.
From this, Agility draws its key conclusion: the answer cannot be left to “the opinion of a private company.” What’s needed are official international standards, ideally written through bodies like ISO. Only then can insurers, regulators, and users judge any new model by the same rules. Agility, for what it’s worth, helped write the forthcoming ISO 25785-1 standard for mobile robots with actively controlled stability and is contributing to an upcoming ANSI/A3 technical report on dynamically stable robots.
Home Humanoids: What This Text Is Really About
On the surface, this is a corporate blog post. In substance, it’s a position paper that pulls expectations back into line with reality. And it deserves to be read as more than just “news from Agility.”
First, the statement about teleoperated demos is a rare moment when a manufacturer openly warns its audience about the viral clips that its own competitors often build reputations on. In effect, Agility is saying: that “autonomous humanoid in a café” you saw on TikTok is most likely being piloted by someone just off-camera. For readers used to judging progress by flashy clips, that’s a significant signal.
Second, the piece rhymes almost perfectly with what former NASA robotics chief Robert Ambrose recently argued in a column for Fortune: the winner won’t be whoever builds the most impressive robot, but whoever first learns to deploy them at scale and safely. Ambrose said it from the outside — as an expert who spent decades building robots for space. Agility is now saying the same thing from the inside — as a company whose robots are already at work on the industrial floors of Schaeffler, Toyota, and other clients. When an independent expert and a major industry player articulate the same idea from opposite directions, it stops being two opinions and starts looking like an emerging consensus.
Third, the promise of “home, eventually” is not skepticism — it’s strategy. Agility is essentially mapping a path to the home that runs through the warehouse: industrial tasks first, data accumulated along the way, costs brought down; then a robot capable of working safely alongside people, without a protective cage; and only at the end, lighter, smaller, more dexterous models that can fit into an apartment, a grocery aisle, or — as the company itself puts it — help with wiring in a cramped attic crawl space.
The takeaway, both for the industry and for the reader, is this: Agility considers the future of “a humanoid partner in every home” entirely real, but insists that it cannot be hurried along with promises. Only with data, standards, and trust. Until then, watch the clips in your feed a little more carefully: what looks like autonomy is, more often than not, still a deceptively disguised trick.