“I Cannot Love Digit 5”: Agility Co-Founder Explains His Robot’s Compromises

“But I cannot love Digit 5.”

That is how Agility Robotics co-founder Jonathan Hurst describes his feelings about the company’s latest humanoid in a new IEEE Spectrum essay. He sees a machine shaped by necessary compromises—and far removed from the graceful robot he wanted to build.

Key Takeaways

Show all

• Hurst describes Digit 5 as a necessary commercial product that falls short of his personal vision.
• Its redesigned legs prioritize lifting, while its planned safety response includes sitting down and disabling its motors.
• Agility expects early access to the Nyobolt-equipped robot in the first half of 2027.

The robot he needs, not the one he dreamed of

Hurst calls Digit 5 the largest, heaviest and most expensive robot Agility will ever build. That last point is his expectation for the company’s future, not a disclosed price comparison.

He describes it as “exactly the robot it needs to be right now rather than the robot I dreamed of.”

The tension gives his account its force: he can defend the engineering decisions without feeling satisfied by their combined result. He ultimately credits customer requirements with exposing gaps in his original vision.

What customers changed

Digit 5’s humanlike leg arrangement addresses a specific problem: the earlier design’s ankles could hit the ground during deep squats. The new configuration lets hip and knee motors contribute to standing up under load.

Safety adds further hardware and weight. Hurst describes a sequence in which an approaching person triggers the robot to stop, put down its payload, sit and cut motor power. That is Agility’s described design approach, not independent confirmation of certification.

The cost of a longer working day

The battery system connects this redesign to Agility’s recently announced Nyobolt partnership. In its October 15 announcement, Agility outlined a target of 90 minutes of operation followed by nine minutes of charging, supporting more than 20 working hours per day.

The company expects early access in the first half of 2027, with availability increasing over the following six to nine months. Those are planned capabilities and rollout dates, rather than results from a broad customer fleet.

For buyers, the useful comparison will be productive hours, handling capacity and deployment costs. A robot that moves impressively in a short demonstration still has to justify its place in an operating facility. Hurst’s account offers a concrete explanation of why optimizing for that job can produce a very different machine.