Avatar Robotics raises $6.5M to turn warehouse shifts into robot-training data

Why it matters

Webb is using paid warehouse work to finance and generate the data for autonomy. The model can reach customers early, but its economics still depend on remote labor and foreign hardware.

Avatar Robotics raised a $6.5 million seed round to expand a warehouse workforce that remains largely human-controlled, giving founder and CEO Colin Webb paying deployments while Avatar works toward greater autonomy. AlleyCorp and Defy.vc led the financing, with participation from Headline and Henry Ford III, according to Business Insider's August 5 report. Avatar did not disclose a valuation.

Webb, 30, has been working toward this experiment since he led his Atlanta high school's robotics club. He graduated from MIT in 2018 and spent a year as an engineer with General Motors' Cruise autonomous-vehicle program. He later co-founded Sauce Technologies, which built pricing and revenue-management software for restaurants and raised over $3.6 million before Linked Eats acquired it in August 2024. (alumni.ronbrown.org)

That sale gave Webb a route back to robotics. He started Avatar in 2025 from his San Francisco apartment using a robot ordered from China. He later demonstrated that he could operate the machine in San Francisco from roughly 2,000 miles away in Orlando, then commissioned stronger grippers after the original components failed under repeated use. (Business Insider)

The experiment became the operating thesis behind Avatar: deploy available hardware before full autonomy is ready, put remote workers behind the controls, and use paid shifts to generate data that can reduce human involvement.

Labor first, autonomy later

Avatar places wheeled, two-armed robots inside warehouses and charges customers a recurring fee rather than selling the hardware. Teleoperators guide the machines with VR headsets and handheld controllers as they pack, pick, sort, kit and move products.

In a Dallas-area pilot shortly before Christmas 2025, one Avatar robot spent an eight-hour shift packing beauty products into pouches beside warehouse employees. The person controlling it was working from an Airbnb a few miles away with a Meta Quest headset. Avatar says its robots have packed nearly 1 million items since a December 2025 commercial launch and are operating across several facilities, though Avatar has not published its robot count, customer count, revenue or exact number of sites. (Business Insider)

Webb told Business Insider that Avatar's recurring fee costs customers less than hiring warehouse employees directly. That pricing rests partly on a distributed contractor workforce. Operators in Mexico, Colombia and the Philippines earn between $5 and $20 an hour depending on location, according to the report. Most currently work from dedicated facilities, although Webb wants people eventually to control robots from home. (Business Insider)

The human-controlled phase gives Avatar a path to revenue before its machines can reliably handle changing products and unstructured warehouse tasks alone. Avatar's platform records what the robot sees and how its human operator responds during each shift. The company uses that information to adapt open-source robotics models, with the goal of having its machines assume more of the work over time.

Webb framed the tradeoff plainly: paying people to perform customer work produces useful data and immediate economic value. "The latter seems like the more valuable path," he told Business Insider. (Business Insider)

Yondu, a warehouse robotics company that also combines machines with remote teleoperation, offers a closer operational comparison than humanoid developers focused on proprietary hardware. Both companies pitch robots that can work within existing warehouse layouts, making the cost and performance of remote operation central to the product.

A seed-stage shortcut through expensive robotics

Avatar is taking a capital-light route compared with developers such as Figure AI and Agility Robotics, which are building proprietary robot platforms. Webb sources Avatar's current machines from China and concentrates Avatar's engineering on teleoperation, fleet software and autonomy.

Avatar can serve warehouses without first designing a proprietary robot platform. The arrangement also leaves a core part of its supply chain outside Webb's control.

Business Insider reported that new Federal Communications Commission restrictions on foreign-made advanced robotic devices could affect Avatar's China-based supply chain and eventually push the company toward US assembly. The timing and direct effect on Avatar remain unclear. (Business Insider)

Webb told Business Insider that Avatar's software can work across different robot platforms. He views the restrictions as a reason to move toward US assembly and, eventually, hardware owned or designed by Avatar. That transition would require a deeper hardware operation than the model that helped Webb reach warehouses quickly. (Business Insider)

The seed round therefore finances two linked tests. Webb must prove that customers will expand from pilots to fleets while Avatar navigates a supply constraint that could narrow its hardware options. He expects individual customers to grow to dozens of robots during 2026 and hundreds in 2027, targets that remain untested without disclosed deployment or revenue figures. (Business Insider)

The operators inside the robots

Avatar's machines remain mostly dependent on people. Its economics currently rely on contractors performing warehouse tasks through hardware that adds capital, connectivity and maintenance costs to the labor equation.

Each operator is also producing data intended to make the machine less dependent on an operator. Webb expects the job to evolve toward supervising larger fleets and intervening when autonomous systems encounter difficult cases, according to Avatar's website.

For now, Avatar needs the humans. Its immediate product is dependable warehouse capacity delivered through a machine, while autonomy remains a development goal. That gives Webb something many robotics founders lack at the seed stage: a service that can be sold before the hardest technical problem has been solved. The next measure is whether those early shifts turn into the dozens of robots per customer that Webb has set as his 2026 target.