The robotics startup that quietly stole the entire show
While the big stages promised humanoid butlers, a ten-person company named Tovera shipped a robot that actually does one job well — and its founder thinks that’s the whole point.

While the big stages promised humanoid butlers, a ten-person company named Tovera shipped a robot that actually does one job well — and its founder thinks that’s the whole point.

Everyone at CES wants to sell you a humanoid. The big booths had them this year — glossy bipeds walking carefully scripted loops behind ropes, gesturing at a future where a robot folds your laundry and minds your kids. They drew cameras. They did not draw me. The most interesting robotics story at the show was happening in a booth a tenth the size, with a machine that doesn’t even have legs.
The company is called Tovera, it has ten employees, and what it brought to CES is a wheeled mobile manipulator — a torso, a single highly capable arm, and a base that rolls — built to do unglamorous warehouse and back-of-house work. No face. No walking. I spent an afternoon there and came away more convinced by it than by anything on the main stages, so I tracked down the founder, Renata Cho, to ask why she went small while everyone else went human-shaped.
Cho is blunt in a way that’s refreshing after three days of pitch decks. “The humanoid demos are a casting call, not a product,” she told me, watching a competitor’s biped shuffle past on a screen. “They’re optimizing for the thing that goes viral. We optimized for the thing a logistics manager will actually sign a purchase order for.”
That thing, in Tovera’s case, is a robot that can pick mixed items off a shelf, move them across an uneven floor, and place them on a packing line — at a claimed 600 picks an hour, with a price she’d only describe as “a fraction of a full humanoid, and we lease it.” When I pushed on the lease number she gave me a range that worked out to roughly the cost of a part-time shift worker, which is exactly the comparison her customers are making.

The arm is the whole company, really. Cho’s background is in surgical robotics, and it shows in how the thing moves — deliberate, compliant, backing off the instant it touches resistance. “Legs are a research problem,” she said. “Hands are a product problem. We put all of our money into the hands and the brain behind them, and we put the robot on wheels because warehouses have floors, not staircases. The minute you commit to walking, you’ve spent your whole budget staying upright.”
What makes Tovera a story rather than just a nice booth is what’s running inside it. The manipulation policies run on-device — a deliberate choice Cho connects to the same theme echoing across the rest of the show floor, where company after company is yanking intelligence out of the cloud and onto the hardware. “We can’t have a robot freeze mid-pick because a network blipped,” she said. “And a warehouse operator doesn’t want a year of their pick data leaving the building. On-device isn’t a feature for us. It’s table stakes.”
“Everyone keeps asking when robots will be general-purpose,” Cho said. “Wrong question. Ask which single job is worth doing perfectly. Then ship that. We’ll get to general later, after the rent’s paid.”
It’s a thesis that runs directly against the maximalist energy of the headliner booths, and Cho knows it. She’s made peace with not being the photogenic one. “We’re not going to be the robot in the news segment,” she said, with the flat confidence of someone who’s done the math. “We’re going to be the robot in the building behind the news segment. There are a lot more buildings than there are news segments.”

She’s also clear-eyed about the parts that aren’t solved. The robot still needs a clean-ish environment and decent lighting; it stumbles on truly chaotic shelving. Battery is a full shift “on a good day,” which in deployment language means they’re shipping spare packs. And ten people is ten people — when I asked how Tovera supports a fleet across a continent, Cho didn’t pretend. “That’s the hard part. That’s the part the funding is for. The robot works. Now we have to prove the company does.”
CES rewards spectacle, and the humanoids will get the clips. But I’ve watched a lot of robots fail to do simple things at a lot of these shows, and Tovera’s quietly did a hard thing — pick, carry, place, repeat — without a handler hovering to rescue it. In our wider show coverage my colleagues kept landing on the same pattern: the standouts this year were the focused ones, the products that did a single job with conviction instead of promising everything. Tovera is the robotics version of exactly that.
Cho put it best as I was leaving, nodding at the biped looping on the screen behind her. “That one’s auditioning for a movie,” she said. “Mine’s looking for a job.” At a show built on auditions, the one looking for a job is the one I can’t stop thinking about.