Let’s start with a story that could have been ripped straight out of an episode of Mike Judge’s Silicon Valley. It’s November 2016 and we’re in a small converted garage office in Pasadena. It’s every “founders in a garage” story you’ve ever heard. Miso Robotics is two months old at the time. Its founders have, at best, a couple thousand dollars in the bank. That and a robot arm that they’ve figuratively pinned their futures on. This is a big day: the day that’s become known as the “first flip.” Miso’s core team of four Caltech engineers have crowded their cramped 400-square-foot office with at least a dozen people. All of them have been promised something remarkable.
John Miller, the owner of a chain of fast food restaurants called CaliBurger, places the first burger on the grill. It starts to sizzle. Miso’s UR-5 robotic arm identifies the burger’s location. Servos whirring and, augmented by a massive game computer for carrying out the processing, it reaches out and scoops up the patty. Then it flips it!
The room goes wild. High-fives are exchanged. Miller, who is Miso’s first customer and an early investor to boot, smiles. But he’s also weirdly silent. There’s a problem. In the team’s desire to build something that no-one’s ever seen before, the Caltech engineers have created a futuristic flipping tool that’s less spatula than it is clamshell; sealing the burger inside like a coffin. It looks great. It looks totally new. It also looks, as Miller quickly confirms, like something that’s impossible to take off and clean. There’s approximately zero chance, he says, that it ever gets certified by the National Sanitation Foundation.
Swapping out the clamshell device for an ordinary food safe spatula doesn’t turn out to be easy, either. In order to have an open-faced spatula perform a perfect flip, the robot needs to have a high twisting speed on both its wrist and joints. The UR-5 robotic arm can’t do that. The Miso Robotics team is forced to throw out all its work so far, find a new robot arm, and rewrite everything from scratch.
All because none of the ultra smart engineers in the room thought about how to clean its smart robot spatula.
Blood, sweat, and burgers
Miso Robotics co-founder and CEO Buck Jordan tells this story with a sense of pride that seems at first misplaced. Mythologizing the early days of a startup is, of course, nothing new. Everyone loves a “founders in a garage” story. But most of these stories are about an early note of triumph; not a slightly embarrassing failure.
In retrospect, though, Jordan is convinced that this lesson — painful as it may have been at the time — turned out to be the best thing that could have happened for Miso. “It saved us so much blood, sweat, and tears,” he told Digital Trends. “It probably saved us a year of development time. What you and I don’t know about burgers would fill a warehouse. There’s a lot of knowledge you need to be a food operator to understand.”
Flippy is the result of the Miso team’s robotics expertise, coupled with that industry-specific knowledge. It’s a burger flipping robot arm that’s equipped with both thermal and regular vision, which grills burgers to order while also advising human collaborators in the kitchen when they need to add cheese or prep buns for serving.
Is building a highly versatile high-tech robotic arm like bringing a laser gun to a fist fight?
But to call Flippy a burger-flipping robot is a bit like calling the iPhone a device that makes phone calls. It is — but it’s also a lot more. With Flippy, the team’s stroke of brilliance was to create a versatile robotic arm solution that can carry out a plethora of kitchen-related tasks. Thanks to a recent redesign, it can now slide effortlessly from workstation to workstation in the kitchen; flipping burgers at one and cooking fries at another.
“Flippy is one form factor that can do everything,” Jordan said. “When McDonald’s or another quick serve restaurant introduces a new chicken tender, buffalo wings or you-name-it, they don’t need to buy a brand new machine. They just program the SOP, the standard operating procedure, for what they want to cook and how they want to cook it.”
To put it in terms more grandiose than a fast food robot possibly warrants, it’s like general intelligence vs. current narrow A.I. Existing machine intelligence is brilliant at doing single tasks exceptionally well. However, it can’t generalize and perform multiple tasks.
“All our competition today have taken this mechanical solution in which they’ve built machines that are very good at doing one thing,” Jordan said. “[Fellow culinary robotics company] Creator makes excellent burgers. They make one of the best burgers I’ve ever had. It’s killer. But don’t ask it to do anything other than that. Don’t ask it to make a chicken burger or deep-fry something. In fact, don’t ask it to make a burger any way other than their way.”
The end to standardization
To play devil’s advocate, you could, of course, ask why it’s necessary to do that. Is building a highly versatile high-tech robotic arm like bringing a laser gun to a fist fight? Can’t a fast food restaurant succeed by just doing one thing really, really well? After all, the recipe used to make a Big Mac has barely changed since 1968, the year it first appeared on the menu. Since then, McDonald’s sells an estimated 550 million units per year. People don’t go to a fast food restaurant to be exposed to new flavors; they go to get something that’s standardized and familiar. Right?
Well, maybe. Except that Jordan believes that the reason we currently expect fast food to be so standardized is because, well, it’s so standardized. Fast food remains an industrial age dream in which mass-produced, pre-packaged products roll off a production line aiming for the lowest common denominator. It’s the old Henry Ford mantra that’s made its way into popular culture as, “You can have any color you want, so long as it’s black.” Only substitute “burger” for “car.” And, unless it’s Japanese Burger King’s limited edition squid ink Kuro Burger, probably not black either. But overall the idea remains the same.
“The reason food is so standardized is because it’s built for an entry-level worker to prepare quickly,” Jordan said. “But what would McDonald’s do if there was a Gordon Ramsay in every one of those stores? Would they still be making the same burger? With Miso, you have the ability to program very complex tasks that require huge amounts of finesse and customization.”
This opens up new possibilities in not just the complexity of menu items, but customer customization, too. Want a Big Mac in which the patty is medium rare? Miso could, at least were it adopted by McDonald’s, “handle that kind of mass customization.”
One inescapable question when it comes to robots like Flippy is what it will mean for human workers. Right now, there are in excess of 3.1 million people in the United States who work as fast food cooks and food preparation workers. Replacing them with a robot that works faster, all-around-the-clock, and wouldn’t dream of taking a vacation or asking for a raise would be a sizable disruption.
Not every job in a fast food kitchen can currently be carried out by Flippy. It can operate the fry station and the grill. Assembly, the job of putting things like lettuce and tomato on a bun, still has to be done by hand. That won’t always have to be the case, though. “It’s something we’ll expand into one day,” Jordan said.
But he disputes the point that robots like Flippy will be used to replace jobs currently being carried out by humans, however. At least, not predominantly. “This is not about replacing people,” he said emphatically. “There’s a massive labor shortage in this area. All of our customers are more concerned about shifts being open than they are replacing workers. They’re not able to fill these shifts. Ten years ago, there was no problem doing this. Today, they’re regularly going short a staff member or two. Sometimes they’re not able to open.”
“I don’t want to build McDonalds. I want to sell to McDonald’s.”
Staffing is a perpetual issue for those people working in the quick serve food industry. These restaurants have a turnover of employees almost as fast as the food they’re preparing. When these roles can be filled, it’s unusual that someone hangs around for long. This means a constant need to retrain people.
“I always think the most exciting exciting opportunities come from massive industries that are in huge amounts of pain,” Jordan said. “No industry is under siege more than quick serve restaurants. The market size is massive, but it’s tough. They’re running at 5 percent margins. Restaurants fail faster than startups.”
The move to automation, he said, is only going to become more pressing as the years go on. “People aren’t cooking at home any more,” he said. “There’s been an absolute explosion of deliveries and delivery services. Millennials order food 3x more often than their parents. We need more commercial chefs in the world. There’s a massive demand on quick serve restaurants to have more stores open. We can’t fill demand today; much less tomorrow.”
While delivery services have embraced new technologies with everything from drone deliveries to delivery robots like those pioneered by Starship Technologies, there hasn’t been that same technological sea change in the kitchen. “There are multi-billion dollar food companies we should be begging to get in front of,” he said. “They’re literally cold-calling us because they want to be able to keep their shifts open.”
The big differentiator
Despite this seemingly rosy outlook, however, the path to robot-driven kitchens of the future hasn’t been straightforward. Zume, a heavily-touted robotic pizza maker and packaging company, went from raising money at a $ 4 billion valuation in November 2019 to shutting down its pizza delivery business and laying off the majority of its workforce at the start of 2020. Meanwhile, automated robotic burger prep company Creator still operates from just one location in San Francisco; a slower rollout than many would have expected. Does this bode poorly for the future of foodie robotics companies? To Jordan, it comes back to that all-important early lesson of the first flip.
“This is the differentiator between us and everyone else today,” he said. “All these other companies, because of the mechanical approach they’ve taken, they rely on building their own restaurant brands. Building a restaurant brand is hard! It’s the toughest industry I can imagine.”
Jordan said that, had it not been for CaliBurger impresario John Miller, “I might’ve gotten cocky and tried to start my own restaurant.” But he’s not convinced this is the way to go. “I don’t want to build McDonalds,” he said. “I want to sell to McDonald’s. McDonald’s already has 48,000 stores. How long would it take me to build 48,000 stores? A lifetime. I think it’s much more likely we can service these massive companies who are banging down our doors with a product, rather than [trying to compete with them].”
When he first began seriously looking at robot arms, they cost between $ 100,000 and $ 300,000. Now they cost between $ 5,000 to $ 8000.
That strategy seems to be paying dividends. Flippy is now found in the Dodger Stadium and Arizona Diamondbacks’ Chase Stadium, as well as two restaurants backed by Cali Group. The firm has just placed an order for another 100 robots, and Jordan said that there are deals with a “couple of large national chains” he’s not yet able to disclose. Because of the myriad abilities of its robot arm, which now takes up less floor space than ever, Flippy can be incorporated into a variety of restaurants.
Miso Robotics is also looking to add new skills all the time. Next year, the company is planning to introduce open software tools that will allow people to program their own skills for the robot so as to integrate it into their kitchens.
“If you own a one-location Mexican joint and want to program the robot to roll your taquito, you’ll be able to do that,” he said.
‘I think this whole thing goes crazy’
Things are helped, Jordan observed, by the decreasing cost of robot arm hardware. What once seemed a science fiction extravagance is now shockingly affordable. Jordan said that, when he first began seriously looking at robot arms, they cost between $ 100,000 and $ 300,000. When Miso started up in 2016, they were between $ 50,000 and $ 60,000. “Now we’re looking at $ 5,000 to $ 8,000 robotic arms that do the same thing,” he said. “There are so many cheap, affordable arms coming out of Shenzhen, and there are some U.S. manufacturers that are trying to aggressively cost down their system.”
Currently, Miso Robotics charges an up-front cost of between $ 20,000 and $ 30,000 to install Flippy. Then there’s a $ 1,500 to $ 2,000 per month fee in the form of software-as-a-service. “It’s highly recurring, very sticky revenue,” he stated. “In fact, it’s the stickiest revenue I would say could exist in the world because if you’re a quick serve restaurant that builds its kitchen around what we’re doing you’re never getting this out of your kitchen.”
2021 is when things get really exciting, though. The announcement hasn’t been made officially, yet, but Jordan believes that it will, by then, be possible to increase monthly subscription fees only slightly, while giving away the robot for free.
“We need to make this thing so affordable that it’s the biggest no-brainer for anyone to adopt,” he said. “Right now, it’s cheap enough given the demand. But tomorrow, when we lower the price even further, I think this thing goes crazy.”