Agility Robotics CEO tells BI how its humanoid robots are getting paid — and warns against empowering them with AI

• Agility Robotics CEO Peggy Johnson says humanoid robots are filling some labor gaps.
• The company’s Digit robots began “working” inside factories this year.
• Agility Robotics is hoping to deploy Digit across industries like grocery, automotive, and pharma.

Robots are coming for our jobs — at least the repetitive, back-breaking jobs humans increasingly don’t want to do.

Peggy Johnson, the Silicon Valley veteran who became chief executive of Agility Robotics earlier this year, said to Business Insider that it’ll soon be “very normal” for humanoid robots to become coworkers with humans across a range of workplaces.

Many factories in the US are struggling to recruit workers amid a labor shortage that Deloitte predicts could cost the economy as much as $1 trillion by 2030. In January this year, there were 622,000 manufacturing jobs that hadn’t been filled, according to data from the Bureau of Labor Statistics.

Enter the robots.

“First in the business-enterprise space because that’s where the need is highest. And then, as Digit learns new skills, it’ll start to be able to go beyond logistics and manufacturing facilities and eventually, somewhere way down the line is consumer robots,” Johnson said in an interview at the Web Summit tech event in Lisbon, Portugal, earlier this month.

Digit is Agility Robotics’ mobile manipulation humanoid robot. It stands at 5′ 9″ and has hands designed to grip and carry objects. Its backward-folding legs allow it to maneuver around a workspace. Digit also has animated LED eyes that act as indicators to its human coworkers to let them know which function it’s about to perform next.

This year, Digit became the first humanoid robot to be “paid” for performing a job. Agility Robotics signed a multiyear deal with GXO Logistics for Digit to be deployed in its Spanx womenswear factories, moving boxes known as totes and placing them onto conveyor belts.

Agility Robotics charges a monthly fee, similar to a software-as-a-service model, which includes the Digit robot, its work cell, and the robot’s operating software.

While Agility Robotics hasn’t disclosed the exact amount its Digit robots are paid, the company has previously said that GXO is estimated to see a return on its investment within two years, based on the equivalent of a human working an hourly rate of $30.

Johnson said that any company that requires material handling — be it pharmaceutical or grocery — could make use of a workforce of Digits.

“Mobile phones started first in the enterprise space because there was an ROI for a salesperson not to stop and find a phone,” Johnson said. “That will happen with robots.”

Amazon began testing Digit in its warehouse operations last year. Ford is looking at how it can deploy Digit with its autonomous vehicle technology to create a “last-mile” delivery service. Most recently, Agility Robotics struck a deal with German automotive and industrial supplier Schaeffler, which also made a minority investment in the company.

Agility Robotics has raised around $178 million in investment to date, a spokesperson said. It competes with the likes of Apptronik, which is working with NASA on humanoid robots, and Boston Dynamics, which has created humanoid robots called Atlas that it says can run and jump over obstacles, as well as perform factory worker tasks.

Agility Robotics’ humanoid robots are currently only permitted to work inside a specific, cordoned-off space separate from human workers. But Johnson said that by mid-2025, the next-generation version of Digit would be able to safely operate around humans. The company is aiming for the new model to be commercially available within 18 to 24 months.

A 2023 Gallup poll found that around a fifth of US workers were worried that their jobs would become obsolete because of technology, up from 15% of workers polled in 2021. Johnson said Agility Robotics hadn’t had pushback from the likes of workers’ unions despite advancements in the number of human-like tasks Digit can perform. Widespread deployment of humanoid robots is still some way off, however.

“I think they also recognize that these are jobs that they haven’t been able to fill,” Johnson said. “We tend to think of it as augmenting humans and not replacing humans — it’s just taking some of the tasks off their plate.”

Hype and misleading marketing videos are ‘not great’ for the robotics industry

While Digit robots are starting to be tested in some workplaces, Johnson said getting them to perform tasks around the home, like folding laundry, will take a while longer.

“A household is a very chaotic environment: At any given moment a child’s ball runs across the room, and dogs run by, there’s things that are in the way,” Johnson said. “Warehouses are much more disciplined.”

Johnson said the data gathered by robots working in warehouses will eventually be used to train consumer robots. But she added that she wants Agility Robotics to focus on demonstrating what its technology can actually perform today — rather than the concept videos used by some of its competitors.

Robotics videos and demos at trade shows and events are often highly choreographed, she said. For instance, Tesla’s humanoid Optimus bots at last month’s robotaxi event were remotely operated by humans behind the scenes.

“The hype, in general, is not great for the industry because people think it’s somehow not here and now,” Johnson said. “My job is to say, no, it is here and now. Humanoids are deployed right now and are getting paid to do work.”

Agility Robotics takes a similarly cautious approach to its application of artificial intelligence, which is deep in the hype stage. Johnson described the company as “AI agnostic,” as it uses various models in reinforcement learning to help fine-tune Digit’s leg movements and help it recognize and carry out various tasks.

“Many companies in the robotics space think, well, now that AI is here, I can just build a complete AI stack. We think that is very dangerous right now,” Johnson said. “The problem is, just asking ChatGPT a question — it doesn’t always answer exactly right. Can you imagine if what it’s telling it to do is move an arm around and these things are human forms, 5′ 9″, 160 pounds? They have a lot of force.”