Agility Robotics has completed a ground-up redesign of its humanoid robot Digit, creating Digit 5, engineered to safely perform warehouse tasks at scale. The new design prioritizes commercial viability and regulatory compliance over the graceful motion founder Jonathan Hurst originally envisioned.
Key Takeaways
- Agility Robotics completed a ground-up redesign of its Digit humanoid robot, creating Digit 5 specifically engineered for commercial warehouse and logistics work rather than iterating on earlier models
- The redesign reflects a fundamental shift from founder Jonathan Hurst's original vision of graceful, elegant bipedal locomotion toward a heavier, bulkier design prioritizing regulatory compliance and operational reliability
- Digit 5 is designed to lift 23-kilogram bins onto high shelves with a 90-minute battery life on a 9-minute charge and work safely alongside human coworkers for up to 20 hours daily
- Agility's Salem, Oregon facility can produce more than 10,000 Digit 5 robots annually, signaling the company's commitment to scaling beyond research deployments to mass production
- Safety and regulatory requirements—including EU CE compliance, UL electrical safety certifications, and FCC electromagnetic limits—emerged as critical barriers to scaling that drove Digit 5's complete redesign
Agility Robotics has launched Digit 5, designed to be the first humanoid robot that can safely and economically do useful work at scale, marking a fundamental shift in how the company approaches humanoid robotics. Rather than iterating incrementally on its existing Digit platform, Agility rebuilt the robot entirely based on a decade of commercial deployment experience and hard lessons learned about what customers actually need.
Founded in 2015 by Jonathan Hurst, Damion Shelton, and Mikhail Jones from Oregon State University's robotics lab, Agility Robotics initially focused on solving a pure physics problem: building bipedal robots that could walk and run with the natural, spring-like motion observed in animals. The company's first project, ATRIAS, demonstrated this principle—a robot that successfully reproduced animal gaits through spring-mass models. That initial success established Agility's technical credibility but also revealed a hard truth about the path to commercialization.
Agility's first commercial product was Cassie, a bipedal research robot with ostrich-like legs designed to maintain balance through active, fluid motion rather than rigid steps. The company sold 10 Cassie units to universities including Harvard, Caltech, the University of Michigan, and UC Berkeley, ultimately saturating the $250,000 research robot market. While Cassie proved the company could build and sell robots, it was not commercially viable as a product for real-world work.
To make the robots useful for actual customers, Agility added a torso and arms to the Cassie platform, creating Digit 1, the company's first mobile manipulation robot. This was a critical step toward creating a machine that could perform practical tasks, not just demonstrate physics principles. However, the company faced a recurring question from potential customers: why build a humanoid robot with legs at all, when simpler single-purpose automation might suffice?
The commercial deployments taught Agility critical lessons about the gap between robotics research and products customers will purchase. In 2019, Digit 1 demonstrated capabilities with Ford Motor Company, carrying packages from autonomous vehicles to front porches. Agility then developed Digit 2 and 3 while working with a global package delivery company on moving and sorting cardboard boxes. Amazon tested Digit 4 on moving and stacking bins, and it became the world's first commercially deployed humanoid at logistics provider GXO. Despite this apparent success, the company faced mounting pressure to prioritize functionality over the engineering ideals that had driven the company's founding.
Founder Jonathan Hurst initially dreamed of building robots that combined locomotion and manipulation to move through the world with graceful, elegant motion—robots that achieved "a mastery of motion, achieved through devices designed to navigate the constraints of physics with elegance." However, deploying Digit robots exposed a hard business reality: customers cared only that robots moved bins, not how they moved. When bins weren't moving, customers were losing money, and Agility had to make continuous improvements to keep deployments operational.
The pressure to transform Digit into a real commercial product intensified when Agility recruited a new executive team with commercialization experience. They insisted that Digit was missing key features essential for a real product. This led to painful compromises between engineering ideals and market realities. Digit 4 EVT (Engineering Validation and Test) was originally intended as a minor upgrade but instead incorporated major changes including larger batteries, system-wide safety and regulatory compliance upgrades, and added mass that dramatically altered the robot's appearance and capabilities. Agility added a head to Digit 4 to optimize antenna and camera placement, which also proved surprisingly useful for human-robot interaction.
For Hurst, watching Digit transform from his vision of a graceful, elegant machine into a bulky, functional workhorse was difficult. He acknowledged that Digit 4 EVT took steps backward from his vision of how humanoids should look and move, as aesthetics and elegance of movement were repeatedly sacrificed in favor of brutal functionality. The design process became "a ruthless triage, with aesthetics and elegance of movement being delayed over and over again in favor of brutal functionality".
Yet commercial deployments provided invaluable learning that would guide the complete redesign. Through Digit 4 EVT's deployments, Agility identified the first 10 use cases Digit would perform and then began designing Digit 5 from scratch based on everything the company had learned.
The redesign prioritizes regulatory and safety requirements that Hurst initially viewed as tangential to his original vision but that proved to be absolute requirements for scaling commercial deployment. Digit 5 must meet European CE regulatory compliance, North American UL electrical-safety certifications, and Federal Communications Commission limits for electromagnetic radiation. Safety emerged as one of the most critical barriers to scaling, and Digit 5 is designed to operate safely next to human coworkers while working 20 hours a day and lifting more than a human can.
Agility's RoboFab facility in Salem, Oregon—a 70,000-square-foot facility—has the capacity to produce more than 10,000 Digit 5 robots per year, indicating the company's commitment to scaling beyond research deployments and early commercial pilots.
Agility's approach to Digit 5's development illustrates a critical tension in commercializing advanced robotics: the engineering drive to solve problems elegantly and the market imperative to solve problems profitably. The company's willingness to abandon years of research into graceful bipedal locomotion in favor of a heavier, bulkier design reflects the practical constraints of scaling humanoid robots from research curiosities to commercial workhorses. Whether this compromise enables Digit 5 to achieve the scale deployments that justify the complete redesign remains to be demonstrated in real-world warehouse and logistics operations.
Sources
- Financial Times: Europe's industrial giants take diverging paths as they ride AI boom
- IEEE Spectrum: The Robot I Didn't Expect to Build
- Financial Times: The battle of the knee replacement robots
- TechCrunch: Laplace Intelligence red-teams robots and self-driving cars
- MarkTechPost: Meet Kyutai REKI
- MachineHours Encyclopedia: AI and Wealth Generation
Frequently Asked Questions
What did the Financial Times report about Europe's industrial companies and AI?
The FT reported on October 18, 2026 that Europe's largest industrial manufacturers are pursuing diverging AI strategies, with some racing to deploy automation and others moving more cautiously.
Why did Agility Robotics redesign its Digit humanoid robot?
According to an IEEE Spectrum first-person account published October 19, 2026, Agility Robotics rebuilt Digit from the ground up rather than iterating on its existing design, reflecting deeper engineering changes ahead of wider deployment.
What does Laplace Intelligence do?
TechCrunch reported that Laplace Intelligence red-teams robots and self-driving cars, stress-testing them before they are deployed in the real world.
What is Kyutai's REKI model?
REKI is a 2.4 billion parameter vision-language-action model from Kyutai, designed to give robots more efficient memory for manipulation tasks, according to MarkTechPost.
How does this connect to job displacement research?
As robotics capability advances faster than labor market tracking, researchers increasingly focus on sector-specific displacement data rather than broad economy-wide estimates, a gap highlighted by Europe's diverging corporate strategies.
Get MachineHours in your inbox
AI compute and the machine economy in your inbox.
