Avatar Robotics

Why this team

CEO Colin Webb and co-founder Nenye Anagbogu met at MIT and have already built and sold a company together: Sauce, an AI pricing tool for restaurant ordering, acquired by Linked Eats in 2024. That tells me they can sell to operators and get to an exit.

What I cannot verify from outside is the depth of robotics experience. The release says the team has worked at Cruise, Apple, Tesla, Intuitive Surgical and Unity, but it does not say who worked where. For a hardware company, the founding engineer's record matters as much as the CEO's.

• Colin WebbFounder and CEOMIT 2018. Co-founded and ran Sauce Technologies, which raised over $3.6M and was acquired by Linked Eats in 2024. restauranttechnologynews.com
• Nenye AnagboguCo-founderMIT 2018. Co-founded Sauce Technologies with Webb. news.mit.edu

What exists today, and what does not yet

Traction on the public record

• More than 900,000 products packed, sorted and shipped in live customer sites since the December 2025 launch. Aug 2026 · robotics247.com
• Two customers described but not named: a global beauty retailer, and a multi-billion-dollar warehouse operator expanding after its pilot. Aug 2026 · pulse2.com
• Deployments are paying; no revenue figure is disclosed. Aug 2026 · runtimewire.com

What I cannot see from outsideRevenue, hourly price, fleet size, gross margin and, above all, how many robots one operator runs today. That ratio is the number this company will be valued on.

Competition, and why they could still win

Reflex is ahead, so the edge cannot be the idea. It has to be speed of deployment and sales. Avatar's founders have sold software to busy operators before, and the plug-and-play pitch, no facility changes, is exactly what a warehouse general manager wants to hear. If they can get to dozens of sites while Reflex is still converting pilots, the larger dataset becomes the moat.

What I would have to believe

1. Teleoperated output can be priced at or near human labor and still leave a gross margin that improves every quarter.
2. The data from paid shifts is good enough to train autonomy for these specific tasks, faster than general robot models arrive from the big labs.
3. Hardware supply survives the FCC Covered List decision, either because the body is not foreign-made or because a compliant one can be sourced at similar cost.
4. Warehouse buyers will take robots-as-a-service contracts from a ten-person company.

Questions for the first call

1. Who makes the robot body, where, and at what landed cost? How does the FCC's July 28 ruling on foreign-made robots affect your next model?
2. How many robots does one operator run today, and what was that number six months ago?
3. What is the hourly price, what does an operator cost you, and what is gross margin per robot-hour?
4. What share of operator actions in your top task could a model already handle, and how do you measure it?
5. Is the beauty retailer paying list price, and what did the warehouse operator's expansion commit to?
6. Where do your operators sit, what is the latency, and what happens when the link drops in the middle of a task?

What gets them to a Series A

• Ten or more paying sites with at least two named, referenceable customers.
• Operator-to-robot ratio moving clearly past one-to-one, with a credible path to one-to-five.
• Gross margin per robot-hour positive and rising for three straight quarters.
• A hardware supply chain that is clear of the FCC Covered List.

When I would walk away

• The robot body is foreign-made and the next model cannot be authorized for sale in the US.
• Operator hours per task stay flat for two quarters, which would mean the data is not compounding.
• Reflex or an incumbent wins the same customers on price.

Risks

• The FCC added foreign-produced mobile and humanoid robots to its Covered List on July 28, 2026; new foreign models now need conditional approval. fenwick.com
• Margins depend on remote human labor until autonomy improves, and the operator ratio is undisclosed. runtimewire.com
• The current robot is rated early: skill 3 of 10, two-finger claws. humanoid.guide
• A direct competitor with the same model has a two-year head start. cbinsights.com