Robot.com Turns Its Delivery Robot Base Toward Workplace Humanoids

Why it matters

Robot.com's move shows how sidewalk robotics operators are trying to turn hard-won deployment infrastructure into a broader labor automation platform. The test is not whether R-noid can demo a task, but whether Chavez can make each new site cheaper to launch than the last.

Robot.com founder and CEO Felipe Chavez is taking the delivery robot operator formerly known as Kiwibot into workplace humanoids, telling Business Insider that Robot.com will launch R-noid, a wheeled humanoid built for repetitive commercial tasks rather than home-assistant theater.

The move is current as of June 22, 2026, but it is not a sudden pivot. Chavez told Business Insider that R-noid grew out of a nearly two-year effort to move beyond campus delivery and into labor automation across food service, logistics, industrial settings and healthcare. Robot.com says R-noid can package orders, load and unload boxes and prep workstations. Robot.com has already commercially deployed fewer than 40 R-noids across about a dozen customers, according to Chavez, including one at Harbor Links Golf Course in New York that helps load food into delivery robots and support order-packing.

That is the important part of the story: Chavez is not trying to sell a sci-fi humanoid from a cold start. He is trying to turn Robot.com's existing delivery footprint, field operations and remote-support stack into a distribution wedge for mobile manipulation.

Chavez's founder story has always been about getting robots into messy public spaces before the technology looked fully ready. In 2018, Berkeley News covered Kiwi as a UC Berkeley startup running small delivery robots on and around campus. Eight years later, Robot.com's bet is that the same operating scars that came from sidewalks, students, support tickets and remote interventions can be reused inside kitchens and loading areas.

The real product is the deployment playbook

Robot.com says R-noid's first job categories will be restaurant assistant, packer, picker, folder and host. On Robot.com's own site, the R-noid line is described as humanoid labor for kitchens, packing lines, order picking and linen folding, with deployment on a customer's floor in weeks. The site also lists other Robot.com lines: R Kiwi for delivery, R Cargo for logistics and R Dog for advertising and inspection.

Robot.com says its legacy delivery network gives it more than hardware. Chavez told Business Insider that Robot.com has 500 robots deployed, most of them delivery robots, and has completed more than 2.5 million tasks. Those are Chavez-supplied figures, but the infrastructure behind them is the strategic claim: operations, maintenance, remote operations, a service desk, data infrastructure and business development already exist. Chavez's blunt summary to Business Insider was: "We already know what it takes to deploy robots."

That is also why R-noid is being pitched as a services-heavy installation, not a boxed product. Robot.com told Business Insider that an R-noid deployment can take eight to 12 weeks. The process starts with a facility visit, task selection and data collection. Chavez said some tasks could require about 50 hours of robot data before the model is fine-tuned and the robot is put on site. He also said Robot.com expects about 70% autonomy during initial deployment, with teleoperations and remote support as core parts of the plan.

That 70% figure matters. It makes Robot.com's pitch more credible and more constrained. R-noid is not being sold as a fully autonomous general worker on day one. It is being sold as an operational system that can get useful work done while Robot.com collects task-specific data, trains models and keeps humans in the loop.

Physical Intelligence gives Robot.com a model partner

Robot.com is working with Physical Intelligence, one of its AI lab partners, to develop custom foundation models for R-noid. Physical Intelligence describes its mission as bringing general-purpose AI into the physical world and building learning algorithms for models that can control robots to do tasks. Its April 2026 research post on pi 0.7 framed the model as a step toward broader generalization across robot platforms, scenes and tasks.

For Chavez, the partnership lets Robot.com avoid pretending that hardware alone is the moat. The harder question for every workplace humanoid company is how quickly each new customer environment becomes repeatable. If every restaurant, warehouse or golf-course kitchen requires a bespoke robotics services project, margins stay trapped in integration work. If Physical Intelligence's models and Robot.com's field data reduce the amount of new data needed per task, Robot.com has a path to something more scalable.

Robot.com has not disclosed the economics that would prove that path. Business Insider reported no pricing, revenue, gross margin, contract-duration or valuation figures. It also did not establish whether the R-noid deployments are paid production contracts, paid pilots or some other commercial arrangement. For now, the measurable footprint is small: fewer than 40 R-noids and about a dozen customers.

Why wheels are winning early

Robot.com's choice of a wheeled humanoid puts it in the practical branch of the humanoid market. The wheeled base gives up stairs and some human-like mobility, but it reduces the burden of balancing a biped while trying to do useful manipulation. That tradeoff is becoming a common pattern among robots aimed at near-term commercial work.

Business Insider pointed to Diligent Robotics, whose Moxi robot works in hospitals, as well as Sunday Robotics and Genesis AI as examples of companies showing wheeled robots. The shared premise is straightforward: many early automation targets do not require legs. They require arms, reliable navigation, workflow integration, uptime and a support model that can survive real customers.

That plays to Chavez's preferred framing. Robot.com's homepage quotes him as saying Robot.com is made of "disciplined builders" focused on scaling proven products. In the R-noid push, that discipline will be measured less by the shape of the robot than by deployment repetition: how many categories Robot.com can automate, how much human remote support each task needs after launch and whether each additional site gets cheaper to stand up.

The delivery robot business gave Robot.com a starting point, but it also exposed the limits of last-mile robotics as a standalone category. Campus and sidewalk delivery can be operationally complex, seasonal, geography-bound and partnership-dependent. Workplace manipulation gives Chavez a larger labor-automation market to chase, but also a tougher bar: customers will compare R-noid not with a novelty delivery bot, but with human labor, traditional automation and doing nothing.

Chavez's line to Business Insider captures the bet: "robots today, not someday." The honest version of that pitch is not that R-noid is finished. It is that Robot.com believes the market will reward founders who can deploy imperfect robots into constrained commercial tasks now, learn from each site and compound the operating data before the general-purpose humanoid race settles.