"One poor man is told to bounce all over the warehouse to find certain items in different colors and in a specific order," Mountz explained to a room full of hardware enthusiasts at the Lemnos Conference in San Francisco on Dec. 6. "We stepped back from the problem and asked ourselves, 'What are we trying to do? We're just trying to put red green & blue together, in that order, in the box, that's it. How can we design a system that stores, moves and sorts all in one?'
Mountz formulated a method centered around automated robots and barcodes, and eventually sought out venture capital to fund his ideas. But during his travels on California's investor-ridden Sand Hill Road in 2002, Mountz admitted having many doors slammed in his face "no less than 50 times." Finally, after a few prototypes of robotic storage and transport systems, Mountz and company received some backing from one particularly excited group -- Mitt Romney's Boston-based Bain Capital -- which helped Mountz launch his hardware company, Kiva Systems, off the ground in 2003.
Kiva originally started working with companies like Staples and Walgreens in 2004, but now, the company lends out its brilliant warehouse system to some of the biggest companies in the US, including Crate & Barrel, Estée Lauder, Dickies, Gap, Timberland, Saks Fifth, Gilt Groupe, Walgreens, Dillard's, Owens & Minor, Office Depot, and Drugstore.com.
With Kiva, Mountz set out with a goal to "turn the warehouse into a parallel processing engine," and in nine short years, his revolutionary technology and state-of-the-art processing system was the key to Kiva's $775 million acquisition by Amazon.com -- a major warehouse user -- in March of this year.
Mountz, who has no problems traveling down Sand Hill Road these days, spoke to the gearheads attending Lemnos Labs' Thursday gathering to talk about the birth of his company, how he and his team solved problems along the way, and how other hardware companies can similarly get noticed in the space. Here's the best of Mountz' keynote speech at the Lemnos Conference, in his own words.
Mountz, on creating Kiva's unique warehouse system:
"What we did at Kiva was we turned the warehouse into a parallel processing engine to put the inventory on mobile shelving pods to bring them to the picker instead of having the picker move all around the inventory."
"There was a process that managed to produce 400 units an hour -- a process called 'carousel picking,' where rotating carousel pods basically move to the front of the room, stop, and a woman finds the lit up box holding the item, and presses a button to make the carousel move again before putting the item in the lit up receptacle.
"This worker in the warehouse has access to every single product in the warehouse without having to walk all over the warehouse. She doesn't have to move a single foot. She can do three to four times as many orders in an hour since she doesn't have to walk around the warehouse. They just find the right item, put it in the box and seal it. It works with bartenders that are scanned every step of the way, which makes orders far more accurate too."
"We realized that it'd be perfect if the robot could just move the shelves around. Then we figured out to maximize each robot: We'd stack the shelves into pods. But since not every pod is going to be used all the time, we realized if we could detach the robot from the pod, that would make things much better."
Mountz, on how the Kiva System actually works:
"There's a centralized server that are having discrete conversations with each of the bots. It looks very sophisticated but each robot is actually very dumb. They don't even know what they're carrying, we just tell them to go lift up a pod and carry it over there and don't ask why.
"The system is very simple. we put stickers on the floor every 40 inches on a grid pattern, and the robots see the stickers and take them into account to drive. There's a segment coordinator, a piece of software in the server, that provents collisions from happening. Until they see and report the next sticker, they stay at the current sticker, which means that no one else will occupy the same sticker. That's what makes everything look like it flows so smoothly, narrowly missing each other."
Mountz, on the interplay of hardware and software in Kiva:
"[We have] a completely mobile inventory. The inventory can walk and talk and comes directly to pack operators. This includes the inventory pods, robotic drive units, operators in pack stations, and the control software.
"The Kiva business has a big piece of software running the entire building. It's a hardware-software play bringing the entire building together, and we've turned the warehouse into an information and bits problem. We can store anything with any algorithm you can dream up on a computer.
"Our system is a massively parralel processing engine, so if somebody steps out and goes to the bathroom, the work doesn't pile up on co-workers. The other constraint we broke with this system of automation is that we can pick from and replenish into the system at the same time. We could now run the building 24/7 without taking time out to replenish the shelving box.
"In software, the whole floor is dynamic and adaptive. We manage the inventory of the pods, we put the faster moving items to the front of the inventory closer to the packers, and the lower moving items farther away. We want to keep robot traffic to the minimum. Everything in our system is designed to optimize the process. Whether you're doing two or 12 or 20 or 200 stations around the perimeter, the system just works. We can literally just pour more equipment into the building and it just sorts itself out."
Mountz, on getting started in the industry:
"I couldn't actually afford an apartment with a garage. I was just coming out of Apple at the time -- I was there for three and a half years -- so I got the idea, started working on it, started doing the initial patent filings for this. But the most important thing for me was to use a white board. My apartment was too small for it so I hung it over the bookshelf in my living room. I wasn't going to read those books anyway. Turns out, there's no faster way to dry up the dating pipeline than putting a white board over your bookshelf.
"But then I started focusing on the business loop. From the product to the team to the customers to the funding. Once you got some dough, you go back to the beginning to make the next verison of the product, and so on. We did about four series of funding, Series A to Series D. We got to where we were on only $33 million in equity-based funding. You can get a lot done with very little cash, as long as you play your cards right.
"The first robot and pod costed about $30,000, but we did it with the product roadmap in mind. You need a product roadmap in mind at all times, just so you know you can solve problems at different times throughout the process instead of trying to do it all at once.
Mountz, on building hardware back then versus today:
"Back then, to create our robots, we used aluminum castings, gearboxes, motors, off-commercially available parts when practical. The whole bot was designed from scratch as a custom product. We could leverage existing tools like gearboxes, we even got one piece that was part of an ice crushing machine. We're using lead acid batteries off the shelf and sensors from sharp -- we think they're actually the ones used for urinals in the bathroom.
"It's easier than ever to get something built [today]. It comes down to three buckets of things: Design tools, ability to protytpe rapidly, and ability to source this sutff and create manufacturing supply chain once you've decided that you're going to be building thousands of these things.
"My perspective goes back to MIT when I graduated with a chemical and engineering degree. Everything today is just quicker and at an effectively lower cost. The internet has made it easier to do stuff, from ordering parts late at night from DigiKey to finding suppliers you never knew existed by typing 'sheet metal vendor New England.' It's fundamentally changed how we did our job."
Mountz, on getting Staples to invest in Kiva Systems in the early days:
"We had seven of these robots, 16 of these pods and two proof stations. The night before Staples came in, we went down to the Staples office supply store, and we bought everything with a Staples logo on it, and we put it all in the system and inventoried it. So when they came to see the system, we were filling and packing their orders. What they told us later, after going top to bottom through the system, the thing that got their attention was putting the 35 lb stack of paper into the system. They couldn't imagine doing it.
"In this competitive situation, our product differentiated them. Staples was opening up a new warehouse every summer, but they were in a battle with Office Depot, Office Max, etc. and they were all selling the same stuff, the same red pens. They were compelled to take a chance with us so Office Max wouldn't take it."
Mountz's advice for people just starting out:
"First, brainstorm the problem. Start with a blank sheet. What is the problem we're trying to solve? What are we trying to do? Try different ideas at the farthest limits, from zero to infinite. Ask yourself, what if we had to build a distribution center in China?
"When you've asked and answered these questions, research the idea, and by all means, go and file your patents. When we had the idea [for Kiva], we felt it was so obvious. Somebody must be doing this. I spent time on the USPTO website and couldn't find anything on mobile robots or mobile shelving, so I wrote and drafted the front of the patents, except the claims, and offered a pair of lawyers $6,000 to write the claims. You want a proof-of-concept in somebody's head.
"Yet, IP isn't worth the paper they're printed on unless you execute. I thought of patent filings were actually just as series of trip wires to prevent others from entering the same space. We thought it was a very intense minefield of issues, but execution was the most important thing.
"You need to know where you're going every step of the way and why. Because we had the product roadmap in mind, we knew that we could solve problems at certain points, rather than trying to do it all at once. we could make quicker decisions and move to market faster."