Robot.com, the Kiwibot spinout, says 500 of its delivery robots have completed 2.5 million tasks. Its new R noid workplace robot is wheeled, not bipedal, and hands the AI to Physical Intelligence's π0.7. The bet is that ops discipline beats chasing the bipedal humanoid.
Robot.com, the San Francisco company that cut its teeth running 500 wheeled delivery robots across college campuses, is making a deliberately unflashy bet on the next phase of workplace automation. Its new robot, R-noid, has no legs and an outsourced brain. Both are the strategy.
The company launched R-noid this week as a wheeled humanoid for industrial, food-service, logistics, and healthcare workplaces, per the company's PR Newswire announcement. Rather than build a bipedal machine that can climb stairs or navigate cluttered homes, Robot.com designed a base that rolls. Rather than train its own foundation model, it handed the cognition to Physical Intelligence, the San Francisco AI lab whose π0.7 model is the silicon that decides what R-noid does when it sees a package, a box, or a workstation.
The bet has two parts and one hidden moat.
The form factor is the first bet. Wheeled humanoids are contrarian in a market dominated by bipedal humanoids from well-capitalized labs. Robot.com CEO Felipe Chavez frames the choice as a deliberate concession, telling Business Insider that the company has decided bipedal performance is a ceiling better-funded rivals will reach first, and that beating them to commercial unit economics matters more.
"The Kiwibot founders and the Robot.com team are a bunch of crazy optimists," Chavez said. "We believe automation is the path forward."
The AI stack is the second bet. Physical Intelligence's π0.7 is a steerable foundation model with emergent capabilities, meaning a single model can be tuned to many robot behaviors rather than retrained per task. For R-noid, packaging orders, loading and unloading boxes, and prepping workstations can be one deployment surface, not nine.
The moat is the one the launch press release does not lead with. Robot.com says its fleet has completed more than 2.5 million deliveries and now runs about 500 robots in active service, per Business Insider's launch report. The company has spent years building the unglamorous infrastructure that bipedal humanoid labs are still trying to figure out: charging, routing, remote ops, and field maintenance at scale.
That infrastructure is what sells to a warehouse manager, not the robot's silhouette. And it is the variable that explains why a delivery startup is making a credible run at workplace humanoids before the better-capitalized general-purpose labs do, as covered in the AOL syndication of the launch.
The honest caveat is that R-noid is a launch announcement, not a shipped product. Robot.com says fewer than 40 R-noids are deployed across about a dozen customers, with one disclosed site at Harbor Links Golf Course. Capability claims are the company's targets, not measured performance. The $10 million raise tied to the Kiwibot era, per Signalbase funding news, predates the pivot in some accounts and is described as powering the automation push in others. The reporting on the launch rests heavily on a single CEO interview and a company press release.
What to watch is whether the next round of customer signatures comes from operators who already run a Robot.com fleet, or whether the company has to win a cold procurement battle against a bipedal humanoid on a plant floor that has never seen a Robot.com robot. The form-factor bet is testable in unit economics. The AI bet is testable when π0.7 ships in real production. The moat is testable when a real operator picks a real robot for a real shift.
The headline writer at Robot.com put it as "No Legs, All Lift to the Bottom Line." Whether that is a tagline or a thesis depends on whether the ops infrastructure survives contact with the workplace.