Berkshire Grey

Berkshire Grey transforms how goods are sorted and moved to people and stores around the world. It does this by combining AI and physical robotics to automate fulfillment, supply chain, and logistics operations. 

In an interview with founder and self-confessed hardcore technologist Tom Wagner, we discuss the company’s vision, how it is giving retailers, manufacturers, and logistics companies a competitive edge, and what the future of AI looks like from his seat.

What inspired you to found Berkshire Grey?

When I left my role as Chief Technology Officer at a publicly traded technology company, I didn’t have a plan — all I knew was that I wanted to use cutting-edge technology in an impactful way.

So, I started inventing things. I spent months working on dozens of ideas and filing numerous patents worldwide, before realizing I was going about it the wrong way – what I needed to do was find a real, commercial problem to solve. I went out to find one. I looked at medical applications and elderly care. I spent time in Detroit looking at manufacturing. I walked the fields of Iowa to talk to farmers.

Along the way, I came across an e-commerce fulfillment operation. It was all manual – enormous numbers of people handling the goods that eventually end up on our doorsteps and store shelves. At that time, tasks like picking things up and packing orders presented difficult technical problems –but they were also right on the edge of what was solvable with AI and robotics. The more we looked at it, the more interested we became. I realized we’d landed on a very special opportunity.

Why did you focus on this sector?

The explosive growth of e-commerce and the increasing demands of modern consumers. Today’s consumers expect more product choices and faster-than-ever delivery times. That has put huge pressure on retail, commerce, and logistics companies to transform the way in which they move goods around, put them into boxes, and get them to our doors at lightning speed.

At the same time, shipping companies and warehouse operations are grappling with an acute shortage of workers to do physically demanding, often poorly compensated jobs like picking, sorting, and packing. There are thousands of open positions they need to fill. If you’re in any doubt, try unloading a complex floor-loaded trailer, where goods are stacked in many forms. That will make you appreciate the need for automation very quickly.

That’s where we come in, filling that gap with intelligent automation to improve efficiency, raise quality, lower prices, and get goods to where they’re needed – smoothly and rapidly.

What do your robots do and how do they work?

Our robots are used in warehouses to help humans with various tasks, like sorting packages of various shapes, sizes, and materials, even when they are all in the same ‘inventory bin’.

Is there a measurable impact from using robots?

Automation removes friction from the process of making and moving goods. Short term, our customers’ costs go down and their operations become much more efficient. Long term, everybody’s costs will go down. People’s dollars will go further and living standards will rise as goods become more available and more affordable.

Some of the benefits we’ve seen through our case studies are a 30% reduction in operational costs and a 50% increase in capacity, using just half the space. Importantly, we also tend to see an increase in Employee Net Promoter Scores because these are hard jobs, and employees are happier doing them with the help of automation. Anything we can do to make it better for these folks is real impact.

When you look to the future, what kind of potential uses of AI do you see within the logistics space?

I think automation will bring a fluidity to the future, where raw materials move through to the manufacturing of subcomponents, then to completed products, then to warehouses, and finally to stores and homes around the world – all fully automated. The cell phone that you order online will be created and almost magically wind up at your door without being touched by a single human hand.

Then there’s the idea that you could reach a point where the device or system anticipates what you're going to need before you’ve even ordered it. Could your refrigerator, for example, detect that you’re running low on iced coffee and then order a replacement?

In this system, things would appear ‘just in time’ in exactly the right quantities, with perfect single-unit granularity. Apply that throughout the whole supply chain, from manufacturing to consumption. You can see how this would drastically minimize waste and resource consumption in terms of inventory, shipping, fuel, and packaging – while maximizing output.

In fact, our systems already monitor their own parts and flag when they need repair or replacing. It’s not been possible on the retail side because of the high cost of a person handling single units, like a pack of gum. But with robots, it works. All these things are feasible, and I believe you’ll start to see them commercially within our lifetimes.

Are there broader applications for your technology?

Our technologies are extremely good at picking up and handling items that they don’t know a priori and figuring out how to work with them. This technology will absolutely move into adjacencies.

For example, within 100 years, I think we will see AI-driven robotics being used to enhance the quality of life of the sick and elderly living at home, helping them with bathing, eating, and living. Even robot butlers. The reason it doesn’t yet exist is because homes are much harder settings than a manufacturing plant or a warehouse. Our homes have different configurations. Our dishwashers are all different. How we live and where we put things is different. Household robots would need to be able to handle a vast array of variables.

More broadly, I think we’ll see a new level of human-AI-robot symbiosis. Whole networks of humans, AI systems, and robots working together to solve problems and perform tasks, each relying on the others’ skills and abilities in a seamless fashion. For example, imagine this blended system working to create a new high high-rise building with robots doing the heavy lifting and difficult, physical, dangerous tasks, while humans and AI work on design and aesthetics. You can see this a little bit today with robot-assisted surgery.

And where do you think AI will take us in the far future, say 300 years from now?

In that time frame, I think there will be advances to go along with AI and robotics – things like programmable matter. In this world, robots, devices – even our clothing, transport and housing might be composed of small elements that are networked with each other and can perform computations and small physical actions. Imagine something as small as a grain of sand, with microscopic physical features, that’s networked with other grains. They’ll work together to form structures, essentially putting networked computing power and AI everywhere.

The future is an exciting place. It’ll take time, energy and investment to bring about those changes. But I’m confident it’ll be miraculous.