Agility Robotics and Digit: The Humanoid That Ships

The Agility Robotics Story

Jonathan Hurst spent years at Oregon State University studying how animals move. His lab, the Dynamic Robotics Laboratory, produced some of the most energy-efficient bipedal walkers ever built. The core research question was simple: if biological systems can walk and run with so little energy, why can't robots? The answer, Hurst found, came down to mechanical design, not just software.

In 2015, Hurst co-founded Agility Robotics with Damion Shelton and Mikhail Jones to commercialize that research. The company's first product, Cassie, shipped in 2016 as a lower-body bipedal research platform. Universities and research institutes bought Cassie as a testbed. In 2022, a Cassie unit set a Guinness World Record for the fastest 100 meters run by a bipedal robot, covering the distance in 24.73 seconds.

Cassie was a proof of concept. Digit was the product. The first complete Digit, released in 2017, added a torso, arms, and a full perception system to the bipedal platform Cassie established. Early Digit units went to research institutions and government labs. Ford Motor Company became the first commercial customer in 2020, exploring Digit for last-mile delivery use cases with autonomous vehicles.

The pivot to industrial logistics came in stages. By 2021, Agility had reoriented Digit toward warehouse work, specifically bulk material handling tasks like tote movement. The real turning point came in March 2023 at ProMat, the supply chain industry's major trade event, when Agility unveiled a new Digit design built explicitly for industrial customers. The message was deliberate: Digit was no longer a lab curiosity. It was built for labor.

The October 2023 announcement of RoboFab was the company signaling real production intent. RoboFab is a dedicated humanoid robot manufacturing facility in Salem, Oregon, the first of its kind in the United States. Agility designed it with peak capacity of 10,000 robots per year using a modular workcell approach that avoids heavy industrial machinery. The facility can scale in sections rather than requiring a complete rebuild to increase throughput.

The Amazon partnership, also announced in late 2023, validated the commercial direction. Amazon's Industrial Innovation Fund led a $150 million Series B, the largest fundraising round Agility had completed. The investment came with intent: bring Digit into Amazon's distribution and fulfillment network, the largest warehouse operation in the United States with roughly 750,000 workers.

Under the Hood: Digit's Design

Digit stands 1.75 meters tall and weighs approximately 65 kilograms. Those numbers put it close to human scale, which is the point: warehouses were built for human bodies. Aisles, conveyor heights, shelf elevations, and doorways are all sized around the average human form. A robot that matches those dimensions can operate in existing facilities without retrofitting.

The arms carry up to 16 kilograms, each with four degrees of freedom. The reach and grip geometry were chosen to handle standard warehouse totes and containers rather than optimized for arbitrary manipulation tasks. Digit runs on a swappable battery system rated for approximately 16 hours of continuous operation, meaning it can work through a full industrial shift without stopping for a charge swap.

On-board compute handles core navigation, perception, and manipulation tasks without cloud dependency. That matters in facilities where network connectivity can be intermittent and where latency in decision-making creates safety risks. Digit can operate autonomously in its primary use cases with the local processor stack.

Illustration

Digit's digitigrade leg architecture. The backward-bending knee joint is the defining mechanical choice that separates Digit from flat-footed humanoid competitors.

The Digitigrade Difference

Most humanoid robots use plantigrade legs, the same configuration as a human, where the heel contacts the ground first and the foot rolls forward. It's intuitive because it mirrors human anatomy, but it requires sophisticated ankle mechanics and creates a relatively wide base of support that doesn't always translate to stability in dynamic environments.

Digit uses digitigrade legs, the same configuration found in birds, dogs, and cats. In a digitigrade stance, the ankle is elevated and the robot contacts the ground on its toes (or the equivalent structure). The result is a spring-loaded leg that handles uneven terrain with less active control effort. Energy is stored and returned through the leg structure passively, reducing motor demand during walking.

This design choice is rooted directly in Hurst's research at OSU. The lab's ATRIAS robot and its descendants demonstrated that passive dynamics, mechanical compliance built into the leg structure rather than controlled by actuators, produces more energy-efficient locomotion than pure motor-driven systems. Digit inherits that lineage.

Agility Arc: The Software Layer

Digit's hardware is paired with Agility Arc, the company's software platform for deployment management, fleet monitoring, and task programming. Arc provides the interface through which operators configure Digit's work assignments, monitor status across multiple units, and update task sequences. The platform is designed for facility managers rather than robotics engineers, reflecting the target customer profile: logistics operators who want robots that work, not robots they need to program.

Humanoid Robot Comparison

Agility Robotics: The Full Picture

Agility Robotics is a private company. It doesn't publish financial results, and precise deployment numbers aren't publicly available. What is known comes from announced partnerships, funding rounds, and regulatory filings.

The founding team reflects the company's academic roots. Jonathan Hurst remains closely involved with the company's technical direction. His co-founders Damion Shelton and Mikhail Jones brought the business and engineering structures needed to move from a university lab to a commercial operation. The company has grown from a small spin-out to a workforce of several hundred, concentrated in Salem, Oregon.

Funding History

RoboFab: Production at Scale

RoboFab is the manufacturing facility Agility opened in Salem, Oregon in late 2023. It's the first dedicated humanoid robot factory in the United States, and its design reflects a specific manufacturing philosophy: skip the heavy industrial machinery that makes traditional robot factories expensive and slow to scale.

Instead, RoboFab uses modular workcells, self-contained production units that can be added as demand grows. The facility's peak annual capacity is 10,000 robots. That number is significant context: most humanoid robot companies haven't shipped 10,000 units total. Building infrastructure to produce that volume reflects confidence in commercial demand at a scale the industry hasn't yet demonstrated.

The facility also signals a strategic bet on domestic production. With supply chain concerns shaping investment decisions across US manufacturing, a Salem-based production line offers customers a domestic supply chain story that overseas alternatives can't match.

What Digit Means for Warehouse Labor

The United States warehouse sector has a structural labor problem. There are more than 500,000 unfilled warehouse jobs, a shortage driven by demographics, working conditions, and competition from other sectors for available workers. Fulfillment centers, especially those running around-the-clock operations, struggle to maintain the headcounts their throughput targets require.

Amazon sits at the center of this dynamic. The company employs roughly 750,000 warehouse workers across its fulfillment network, making it one of the largest employers in the country. Recruiting, training, and retaining that workforce at the volumes required for Prime-speed delivery is a constant operational challenge. Agility Robotics is positioning Digit as a partial answer to that challenge.

Illustration

The cooperative future: Digit is designed to work alongside human coworkers rather than replace entire workflows. The 2026 cooperatively-safe certification targets regulated shared workspaces.

The RaaS Model

The Robot-as-a-Service model Agility uses for its GXO Logistics deployment is a deliberate response to the economics of robotics adoption. Buying humanoid robots outright requires capital expenditure that most logistics operators can't justify without proven ROI data, and that data doesn't exist yet for humanoid robots specifically.

Under RaaS, customers pay a per-robot-per-hour rate. The burden of capital investment stays with Agility. Customers get flexible capacity: scale up during peak seasons, scale back in slower periods, without sitting on depreciating hardware assets. Agility carries the utilization risk, but also captures the recurring revenue that makes the business model more predictable over time.

The GXO contract at the Spanx facility outside Atlanta was the first humanoid RaaS deal announced in the industry. Digit moved totes in that facility under a structured deployment, not a demo, meaning it was integrated into actual production workflows with real throughput expectations.

Safety Certification: A Practical Barrier Cleared

In 2025, Digit became the first humanoid robot to earn both FCC approval and NRTL certification (National Recognized Testing Laboratory, the safety standard required for equipment used in US workplaces). These certifications aren't marketing milestones. They're operational prerequisites.

NRTL certification means employers can deploy Digit in regulated work environments without creating OSHA compliance problems. FCC approval covers the wireless communications systems Digit uses for fleet management. Together, they clear the path for enterprise customers who couldn't deploy uncertified equipment even if they wanted to. Agility is the only humanoid company to have cleared both bars, and the lead it holds on this front is not easily closed by competitors still in the certification pipeline.

What's Coming Next

The 2026 target for cooperatively-safe certification would be the next major milestone after 2025's regulatory wins. Cooperatively-safe status, distinct from the existing NRTL certification, would allow Digit to operate in closer proximity to human workers under formal safety frameworks, expanding the range of tasks it can perform without physical barriers or exclusion zones.

The Amazon deployment trajectory is the largest open variable. Amazon has not publicly disclosed the scale or timeline of Digit's rollout within its fulfillment network. The partnership was announced in late 2023, and initial pilots have been underway since. A broader deployment at Amazon's scale would represent a step-change for the entire humanoid robotics sector, not just Agility.

RoboFab's production ramp is the supply-side constraint to watch. Building 10,000 humanoid robots per year requires consistent component supply chains, trained assembly workers, and quality control systems that don't yet have a decade of refinement behind them. How quickly Agility can close the gap between RoboFab's peak capacity and actual production volume will shape the company's ability to fulfill large enterprise contracts.

On the software side, Agility Arc's evolution determines how quickly new customers can deploy Digit without custom integration work. The company's long-term value proposition depends on making the software layer generic enough that a new logistics customer can onboard Digit in weeks rather than months. That's a software problem more than a hardware problem, and it's where much of the company's development resources are now focused.

Frequently Asked Questions