Digit at Work: Agility Robotics' Humanoid Passes the Commercial Test

At a Spanx distribution center outside Atlanta, a 175-centimeter robot walks between autonomous mobile robots and a conveyor belt. It picks up a tote, turns, places it. Repeats. The shift lasts eight hours. The next shift, same thing.

This is not a demo. GXO pays Agility Robotics for this under a multi-year Robots-as-a-Service agreement. It has moved more than 100,000 totes since June 2024. Digit is the first bipedal humanoid in the world generating revenue from paying commercial customers.

As we wrote in July, Agility Robotics went public through a merger with Churchill Capital Corp XI. The SPAC deal gave the company a public-market footprint. The story that defines this moment is Digit — a robot that has been working a factory shift every day for two years.

From Oregon Lab to Factory Floor

Agility Robotics was founded in 2015 by Damion Shelton and Jonathan Hurst, both robotics researchers from Oregon State University. For the first seven years, the robot existed inside the valley of prototypes — conference demos, pilot runs, press releases promising a commercial future that kept receding.

The change happened in June 2024. GXO Logistics signed a multi-year commercial agreement to deploy a small fleet of Digit robots at its Spanx facility in Flowery Branch, Georgia. Not a pilot. Not a test. A paid, multi-year RaaS contract. The Robot Report called it the industry's first formal commercial deployment of a humanoid robot.

Since then, it has accumulated more verified production hours than every other Western humanoid combined.

The Numbers Behind the Robot

Digit stands 175 cm, weighs 64 kg, and lifts 16 kg per cycle. Its bird-inspired legs let it walk stairs, tight aisles, and uneven floors. Wheeled robots fail in those environments. It runs on Agility's cloud platform Arc, which handles fleet orchestration, data collection, and over-the-air updates.

At GXO, Digit unloads totes from autonomous mobile robots and feeds them onto conveyors. The facility runs three shifts. The robot runs with them. A single Digit costs roughly $250,000 under the RaaS model, with an operating cost of $10-25 per hour. Agility's long-term target is $2-3 per hour — competitive with manual labor in US logistics.

Four Customers, One Robot

The customer list is a cross-section of industrial America. GXO in logistics. Schaeffler in automotive components. Toyota Motor Manufacturing Canada in vehicle assembly. Mercado Libre in e-commerce fulfillment. Each deployment started as a pilot, de-risked through on-site validation, and converted to a commercial agreement.

Toyota's path shows the pattern. The automaker tested Digit for a year across three phases at its Woodstock, Ontario plant. In February 2026, it signed an agreement to deploy seven Digit robots for tote handling in RAV4 production logistics. "After evaluating a number of robots, we are excited to deploy Digit to improve the team member experience," said Tim Hollander, president of TMMC.

Every current deployment is single-site. Schaeffler's "100 plants by 2030" is an aspiration, not a schedule. GXO has not announced a second site. But intra-site scaling is happening — Toyota is moving from pilot to 7 units at one plant, which is the pattern that matters at this stage of the market.

The RaaS Model as a Moat

Agility sells the robot as a service, not a capital purchase. This is strategically important for two reasons. First, it lowers the adoption barrier for customers who are skeptical about humanoid ROI — they pay for uptime, not hardware. Second, it gives Agility direct operational data from every deployed unit, creating a feedback loop that competitor hardware-only vendors cannot replicate.

The RaaS model also means Agility carries the maintenance, software updates, and safety compliance costs. This is capital-intensive — Agility raised roughly $400 million in 2025 at a $2 billion valuation — but it creates switching costs. A customer running Digit through Arc cannot easily swap in a competing robot without rebuilding the orchestration layer.

Safety Is the Real Constraint

Every Digit currently operates inside a cage or segregated zone. The robot cannot yet detect humans, which limits its addressable use cases to environments where physical separation is feasible. CEO Peggy Johnson targets a cooperatively safe Digit v5 for early 2027 — meaning the robot could work alongside people without barriers.

The ISO 25785 safety standard for dynamically stable bipedal robots is three to four years from ratification. Agility passed the first OSHA-recognized NRTL safety field inspection at a live customer site in November 2025 — a milestone, but not the same as certification. Until cooperative safety is solved, Digit's market is confined to segregated zones in warehouses and factories.

This constraint is not unique to Agility. Every humanoid vendor faces the same safety gap. The difference is that Agility is the only one with real operating data to inform the standard's development.

What Digit Proves About the Market

The humanoid robotics industry has raised more than $8 billion in startup funding since 2023. Most of that capital is betting on a future where general-purpose humanoids replace human labor across manufacturing, logistics, and services. Its deployment record offers the first real data point on whether that thesis holds.

The evidence so far: humanoids work in narrow, repetitive logistics tasks within segregated zones. They deliver concrete numbers: 99% placement accuracy at BMW, 100,000 totes at GXO. They do not yet work alongside humans, do not generalize across tasks, and require significant infrastructure investment per deployment.

This is what early-stage industrial automation looks like. The first CNC machines also ran inside cages.

Sources