In the second installment of IEEE RAS’ series on commercialization, we spoke with Damion Shelton, co-founder of Agility Robotics and associate vice president and executive director of Carnegie Mellon University’s Swartz Center for Entrepreneurship.
Shelton discussed Agility’s choice of an initial market for Digit, lessons from mentoring founders, and how universities can help researchers prepare for commercialization.
The interview has been edited for length and clarity.
How did you decide on the first commercial application for Digit?
Damion Shelton: When we initially started the company, we knew that we had to go through an early research phase. That was the Cassie robot, the predecessor to Digit. We always joked with our early investors that people would ask us what Cassie was good at, and our answer was that it was good at being a robot. There wasn’t really an end use case for Cassie.
With Digit, it was always focused on bulk material handling. But that raises a question: what bulk material handling application are you actually targeting?
We had a partnership with Ford Motor Company. They had a really strong self-driving vehicle research program, and we were interested in exploring last-mile delivery with them. Imagine a future where you’re using self-driving vehicles for delivery. The challenge is how you maintain that same quality of service without dumping the package in someone’s driveway. The partnership of a humanoid robot with a last-mile delivery vehicle was quite compelling.
It was also a structural weakness. It takes a hard deep-tech problem, humanoid robots, and makes it dependent on another hard deep-tech problem, self-driving vehicles. It ended up not being a great first use case for that reason.
That’s how we got focused on internal warehouse tasks. It’s very similar in terms of the mechanical needs of the robot, but it removes the dependency on external factors because you’re inside a closed environment. That also provides benefits from the safety and market approach standpoints.
The other realization was that the warehouse sector is just huge. There’s an enormous need for automation driven by labor shortages. So, it’s a great market to be in, and something our robots were demonstrably able to do.
What are the most important business lessons you have taken from your experience at Agility?
Damion Shelton: The most important thing is to build something that’s useful to someone rapidly, as opposed to having it be an indefinite science project. Get something out the door as rapidly as possible, even if you’re a hardware company, and learn from having it in the field.
Although the Cassie robots were targeted at research use, seven months into the history of Agility we were having to ship products to customers, handle technical support in the field, handle customer feature requests, and so on. It was a limited market, but nonetheless we were forced to ship and maintain product rapidly.
Agility has done that with almost all of the versions of the robot that it’s had. We actually get them out into the world, doing real stuff. You can learn an awful lot from interacting with customers and partners.
Was Cassie also good for Agility on the PR side? It seemed to capture the imagination.
Damion Shelton: Yes. No question. Cassie came across as organic. It was dead silent when you stood next to it, so it was not some big hissing, whirring industrial machine. It had a lot of charisma.
Cassie also had really good leg design and biological-looking motion, even if that motion was really derived from first-principles control theory and mechanical design.
Overall, this created a very charismatic, fun-to-be-around robot that had an excess of personality. People just identified with it.
In your mentoring work, what is the most common issue that you have to guide founders through?
Damion Shelton: Everybody who’s a technical founder is fundamentally a technology enthusiast, and that can confuse you as to what your customers are looking for.
Melonee Wise had a saying that I just love: If you are a warehouse manager and you are shopping for something, what you’re shopping for is a Star Trek teleporter. The reason you settle for a robot is that you were shopping for a teleporter, and you couldn’t find one.
Really, all customers want is to move stuff from point A to point B. They’re shopping for a solution that does that, and that’s it. If you have the best solution to their problem, then they’ll buy it. If you don’t, then they won’t.
That gets overlooked by technical founders with some frequency. You really have to offer a solution to someone. It requires a lot of actual research into the field to understand where the market opportunity is. But that opportunity is going to take the form of solving a customer’s problem.
For technology nerds like us, it’s sufficient that something is a cool piece of technology. For a consumer of a product, it certainly helps if it’s fun and charismatic and interesting to look at. But at the end of the day, you’re buying it to solve a problem.
What do you see as the biggest gap in how universities help researchers commercialize robotics, and how do you hope to address it in your new role at CMU?
Damion Shelton: On the scale of university budgets, a million dollars is a huge expense to put into a company. On the flip side, these days, a million dollars is a very small investment round.
There’s not really anything universities can do financially to make a company succeed or fail, but universities have a superpower. They can keep moving forward with technology development while the company is still in an incubation stage.
That’s what Agility did.
We started talking about proto-Agility in 2013. We had conversations with DARPA and the university before receiving a grant that ultimately resulted in Cassie being given to the university to keep everybody aware and on trajectory.
Public-private partnerships are the right way to produce tech commercialization. The problem universities have had historically is wanting to launch something into the world without considering the entire arc of the tech-transfer process.
The arc can be traced from foundational research through company formation, IP licensing, early-stage funding, later-stage funding, and finally some sort of long-term disposition, whether it’s continued operations, a sale or an IPO.
Planning all of that out can be done for free before the company really exists. Universities can be very active participants in that early ideation and formation phase, so that when the company finally gets incorporated and has employees, it’s a very capable offering to the market for fundraising. That’s opposed to getting to the point where you have a patent and just hoping a licensing opportunity comes along.
Do you see a role for the university in creating a broader ecosystem for commercialization and networking?
Damion Shelton: Certainly. Imagine 10 or 15 venture firms all committed to having a junior associate on campus for two days a month. Suddenly, you can have conversations on a weekly basis across the venture ecosystem. Engagement can occur earlier and more regularly.
The role of the Swartz Center at CMU is to break down some of the barriers to commercialization and to act as a clearinghouse between the worlds of venture capital and research.
Consider the following failure mode: A faculty member or grad student works really hard on a project and decides to commercialize it. The first time they ever interact with a venture firm is when they fly out to Sand Hill Road, get in front of a VC, pitch, and it goes terribly. They’re told “No. That’s not how the system can or should work.”
They should have been talking with that same venture firm for the prior two years, while this was still an academic idea. That way, it’s much a more gradual progression through the process of getting yourself venture-ready.
Don’t turn the process into a high-stakes, sudden death where you find out very late that this isn’t packaged in a way that’s venture-ready.
Conversely, an outside investor could say, “This is super promising. We’ve done market research that says this clearly needs to exist. We’re willing to sign up to fund that now.”
That was exactly what we did with our seed investors: “We’re working on this. It’ll be at this point a year from now. Would you like to commit to funding the technology?” And they said yes.
How has the robotics commercialization landscape changed since you founded Agility?
Damion Shelton: There’s so much more money available now.
The AI ecosystem has overturned a lot of assumptions around capital flows, but also the efficiency of very small early-stage startups. You’ll see startups with lower headcounts but much larger spending on things like inference and compute.
Some of that is good, some of that is bad. But it’s netting out to be a really interesting landscape where you have proportionately smaller startups raising larger rounds with fewer human employees and deploying capital faster.
That’s a very new landscape. I don’t have a prediction about where it’s going, other than to say that it’s definitely different from where we were when Agility was getting started.
What do you see as the biggest opportunities in robotics and automation commercialization today?
Damion Shelton: That hasn’t changed a ton in terms of demand backlog. The classic robotics areas, the dull, dirty, dangerous work, still exist. Exciting stuff is being done with infrastructure and warehousing.
I’m a little less enthusiastic about home robots for the time being. They have a much stricter safety profile, and I think the economics are a little less clear. For the shorter term, the classic dull, dirty, dangerous stuff that has been the hallmark of robotics deployments for the last 40 or 50 years remains the best opportunity.
I was just on a call with folks here at CMU in the energy group. The buildout of the electrical grid, pipelines, power production, wind turbines, solar, natural gas, all that is driving a huge push for skilled trade labor. There’s not enough of that either.
Do you have a final message for the founders of the future?
Damion Shelton: As a founder, it’s incredibly important to do something you find fun and rewarding. It’s an incredibly hard lifestyle to pick, and you have to be aware of that going in.
The most boring reason to start a company would be that you thought you were going to get rich. Honestly, there are a lot of easier ways to make money than starting a small company and trying to scale it.
Find something you’re passionate about and do that. Then figure out how to keep that going for yourself, where you’re doing it because it’s cool, but people are paying you because it’s valuable. That’s the entire trick.
If you fail at either of those things, you’ll either burn yourself out because you’re doing something you hate, or you won’t be able to sell it, in which case the company isn’t viable.
So, figure out something that’s sustainable, exciting, and meets your intellectual goals. Then figure out how to get somebody to pay you for it. That’s the difference between a company and a lifestyle choice.
Emmet Cole, Science Communicator
