It’s rare that a project survives the death of the innovation lab that hatched it.
But an idea that was originally dubbed Project Tricorder, born five years ago in the Hong Kong offices of the multinational grocer Tesco, survived not one but two near-death experiences. The notion was to develop a handheld scanner that could analyze the freshness of a piece of fruit or fish, much the same way that the Tricorder in Star Trek could analyze the health of living things.
Two former Tesco employees, Greg Shewmaker and Brent Overcash, took the idea with them when they joined Target in 2015, and continued to develop it under a new name: Illuminate. But when the Minneapolis retailer curtailed much of its innovation investment in 2017, the food analyzer was once again an orphan.
Shewmaker and Overcash weren’t ready to shelve it and move on to something else. They had met with investors who might be willing to provide capital. They had met with a potential partner—the lab equipment maker PerkinElmer—that might provide other support.
As spring turned to summer in 2017, Shewmaker and Overcash wondered if they might be able to turn Illuminate into a standalone venture. Could a pair of intrapreneurs succeed as entrepreneurs?
Shifting Food Analysis from Art to Science
Even at a high-end sushi restaurant, as much as 60 percent of the “tuna” you order is actually some other fish, says Overcash. And “extra virgin olive oil” may be blended with other kinds of less-expensive oils, like safflower or chestnut. The average apple you pluck out of a grocery store bin has about half the nutritional value that the United States Department of Agriculture standards say it should have.
“We think you deserve to know that,” Shewmaker says. But there are limitations in the way most grocers sample the food items moving through their supply chain. Quality control often consists of cutting open a piece of fruit to smell or taste it, or using a device called a refractometer to measure its sugar content—what’s called a Brix test. After spending time in various grocery distribution centers watching a single sample stand for the quality of a palette-load of produce, Shewmaker says he concluded that “it’s a total art.” What if you could develop a technology that could make it more of a science—and allow grocers to test more of what was coming into their stores?
When Shewmaker and Overcash initially began developing the concept, they were running a group in Hong Kong called Tesco Advanced Product, an innovation lab far away from corporate headquarters in the UK.
“We could partner directly with a giant manufacturing base in Asia,” Shewmaker says. “We could move at Asia speed, versus London speed.” One of the group’s projects was creating an inexpensive tablet computer called the Hudl that Tesco sold through its 6,800 locations. Tesco sold several million Hudls, and the next project for Tesco Advanced Product was supposed to be a smartphone. Shewmaker and Overcash were exploring ways to make it “sexy and exciting beyond just the standard Android device.” One possibility they tossed around was building some kind of food-analyzing technology into the phone, or a phone case, so that consumers could run their own quick scan on a peach or persimmon while in a Tesco store.
“It was a concept—just videos and PowerPoint decks—at that point,” says Overcash. They were starting to make the pitch for funding, but they hadn’t started to develop the underlying technology that would be required.
Tesco’s executive team seemed supportive. But suddenly, in September 2014, the company revealed it had overstated its profits in the first half of that year by more than $300 million. Tesco’s stock lost almost half its value in the wake of the revelations, and several executives were fired. Britain’s Financial Conduct Authority launched an investigation. The Tesco Advanced Product group didn’t get shut down abruptly, Shewmaker says, but rather “faded away, with the team leaving to pursue other things, and the organization focusing on damage control and core ops.”
The Target Era
As Tesco was grappling with the accounting scandal, a recruiter had come to Shewmaker with an opportunity: join Target as an entrepreneur-in-residence, as part of a new program that was being launched in 2015 to help the company incubate new ideas. He decided to take the job, and move to Minneapolis. He’d be working closely with Target’s Chief Strategy and Innovation Officer, Casey Carl. And he’d have access to an initial “seed fund” of $1 million.
“Rather than go and dabble with the million bucks—take some Post-it Notes, build some cardboard prototypes,” Shewmaker says, he opted to “go all in” on continuing to pursue the germ of the Project Tricorder idea. He found an engineering firm in Florida that focused on building measurement instruments, Ocean Insight, and tasked it with building a prototype of the food analyzer. The goal was to have something to demonstrate to Target’s board of directors within three months.
Eleven weeks later, Shewmaker had a working prototype. He showed how it could assess the quality and authenticity of meat, apples, and wine. “It was sci-fi happening on the top floor of Target headquarters,” he says. In a complicated grocery supply chain like Target’s, executives felt it might help them pinpoint where problems were occurring, and make sure high-quality food was ending up on shelves. The project had been rebranded from Project Tricorder to Illuminate.
The finished Illuminate device, they posited, would be something that could be used in Target, and sold to others throughout the food supply chain. “As an entrepreneur-in-residence, the idea was to create a business that could be separate, and test it inside of Target,” Shewmaker says. “They’d get first dibs, but it would be released to the world eventually.”
The demo in the fall of 2015, Shewmaker says, helped unlock more funding. “It was the genesis,” he says, of a bigger initiative at Target called Food+Future, which involved collaborations with the design firm IDEO and MIT’s Media Lab. They set up a basement innovation lab in Cambridge, Mass., just a few blocks from the edge of the MIT campus and the local IDEO office. Overcash moved from Hong Kong to the Boston area. The lab was home not just to the Illuminate effort, but several other projects—some being developed by MIT researchers and students, and some proposed by entrepreneurs that had come to Target-sponsored events focused on three key themes: understanding your food, accessing better food, and trusting your food.
Shewmaker and Overcash now had access to Target stores and warehouses, where company staffers would test various versions of the Illuminate technology. One device, about the size of a microwave oven and dubbed the “Super Bench,” was built to compare the accuracy of spectrometers available at different price points. (A spectrometer uses light to determine the chemical composition of an item—it can be used to determine whether a 14-carat ring is actually 14 carats, or whether a piece of tuna is actually tuna.) Inside it were a $50,000 spectrometer, and also one that cost just $300. That let them compare the quality of data they were getting from each one.
“We wanted to prove we could do one million scans” of food items in real-world environments, Overcash says. As they progressed toward that goal, they were also realizing that there was no “gold standard data” to compare food items to. What was the chemical make-up of a perfect pear, or a ready-to-eat banana? A lot of the data they wanted was in disparate nutritional databases. Much of it was very old, and based on a small sample sizes. Some of the data was simply “unusable with modern-day technology,” Shewmaker says. He and Overcash were slowly realizing that there was no one building a “common source” of data about food—a reference not unlike a pricing book for antique collectors, or a field guide for birders.
But before they could assemble that kind of dataset, or turn their bigger devices into finished handheld “ray guns” that could be deployed broadly, there were worrisome signs in the company’s core business. Target had a challenging holiday season in 2016, with sales and store visits dropping. “Overall, the season simply was not up to our expectations,” Target CEO Brian Cornell said, in an interview on the company’s website. The company lowered its profit forecasts.
By February 2017, Target’s hometown newspaper, the Minneapolis Star-Tribune was reporting that a Store of the Future project was being shelved. By April 2017, the company’s Chief Strategy and Innovation Officer, Casey Carl, was on the way out. By May, Shewmaker got the news that his group wasn’t going to survive.
“It was baby and bathwater,” he says. “It was about focusing on the core, and investing in the stores.” Shewmaker says he “understood it with my corporate hat on. It was the right decision for them… But I was disappointed.”
The same day that Shewmaker and Overcash got the news, they “immediately shifted gears,” Overcash recalls. “We were thinking, how do we survive after the fact?”
Finding a New Home
Almost all of the staffers at the Food+Future lab in Cambridge scattered after the news came. Shewmaker says he didn’t want to make false promises about whether they might be able to find a new source of funding, and “I wanted everyone to make sure they were focused on themselves first.”
One of the companies that Shewmaker had encountered through Target’s sponsorship of the MIT Media Lab was LKK Health Products Group, a Hong Kong-based food ingredient company. Their executive team had visited the Food+ Future lab. “They’d seen the demo many times,” Shewmaker says. “I called them up, and they basically said, ‘We’re in.’ It was a one-phone-call pitch, and we were off and running.” Shewmaker says that the investing division of LKK, Happiness Capital, agreed to put $2 million into a new entity that would keep developing the Illuminate concept.
Target wanted to see the project survive, Shewmaker says. Their sole stipulation: just don’t keep calling it Illuminate. They were “worried about the association,” he says. “If this thing is wildly successful, they look really stupid because they killed innovation. Or if we do something dumb, and are a failure,” that could also reflect badly on the retailer. But Shewmaker says they were generally supportive.
The new name that Shewmaker chose: TeakOrigin. “Brent and I were scrambling to set the company up, and trying to think of a name,” he explains. Shewmaker owns a house in Woodstock, Vermont, and TeakOrigin is intended to be a synonym (teak being a kind of wood, stock being a place of origin). “I thought the name was dumb, but it stuck,” he says.
Overcash had had some early conversations with PerkinElmer, a maker of lab instruments—including spectrometers and other analytical devices. When it came time to try to set up Illuminate as its own venture, he got back in touch. “We needed lab space,” Overcash says, “and they had some empty, brand new lab space” in the first floor of PerkinElmer headquarters in the Boston suburbs. The company offered TeakOrigin the lab space for free—plus some of its high-end lab instruments—without asking for equity or rent in return. The larger company hoped that the startup would help test some of the new equipment it was making, and perhaps collaborate on some new, lower-cost instruments especially geared to the food industry.
Just three employees—Shewmaker, Overcash, and Chief Scientist Ellen Miseo—made the leap from Target, and the Illuminate era, to TeakOrigin.
From Hardware to Software
While at Target, Shewmaker and Overcash had spent much of their time—and budget—building proof-of-concept devices. But as a newly-independent company, they started to shift their focus from hardware to software. Other companies, they’d observed, were beginning to produce inexpensive spectrometers. There was even one for sale on the crowdfunding site Kickstarter, for $300.
There were also companies trying to apply artificial intelligence to the problem of analyzing food: just look at enough oranges, and eventually the software will learn the characteristics of a perfect orange. But he and Overcash felt that there was still a need for a “common source of data—otherwise these AIs and devices will never work in the real world,” Shewmaker says.
Gathering that data became TeakOrigin’s sole focus. One of its datasets describes the sort of “Platonic ideal” of different kinds of food, starting with popular items like apples, bananas, avocados, and grapes. Another dataset assembles data collected from the field—the aisles of major grocery chains—so that grocers can see where they stand relative to competitors. An early test analyzed strawberries and found that Walmart was offering the top quality berries, and Costco was bringing up the rear. “Walmart could go out and say, ‘We’re giving you the same quality, at a lower price, than everyone else,'” Shewmaker says.
At a countertop in the company’s lab, Dan Accetta, a data scientist at TeakOrigin, is wielding a $1500 hand-held scanner made by a Taiwanese company called InnoSpectra, connected by Bluetooth to a laptop. The subject of scrutiny this morning is an appetizing-looking Fuji apple. In 15 seconds, the scanner takes several hundred scans of the apple, and sends them to be interpreted on the laptop using TeakOrigin’s software and database. The score? This apple is a 7, on a scale of 1-10. It gets decent marks for vitamin C, but for a Fuji, it’s kind of low on antioxidants.
Accetta is one of nine employees at TeakOrigin. The company has raised just over $5 million in funding from investors. Its plan is to launch what Shewmaker calls the Food Quality Index in November of 2019—a way for one retailer to compare the quality of its produce to competitors. Then, in January of 2020, the company will be having its official “coming out party,” he says, as part of the National Retail Federation’s annual trade show in New York.
Life at a Startup
Sitting at a table in a Spartan café area in the PerkinElmer headquarters building, Shewmaker reflects on the differences between hustling to make a startup successful, and getting a new venture going inside a more established company.
“Starting a company is the dumbest possible business model you could dream of,” he says. “You have no resources, and no scale. The best place to do something new would be inside a massive organization with all those resources, but culturally, it’s very hard to do.”
One key difference, Overcash interjects, is that as a venture capital-funded startup, “we’re better able to control and predict our destiny. In Target, you could walk in, and it’s over one day. We can at least see it’s coming, because we control our own money. With these corporate innovation initiatives, you don’t control your own destiny. You’re a budget code.”
Thinking back on the Target experience, Shewmaker muses that perhaps he should’ve spent more time at company headquarters in Minneapolis, “managing those expectations and relationships.”
Would it have made a difference? Probably not, he admits.
These days, Shewmaker doesn’t have to worry as much about optics or executive expectations inside a multi-billion company. But he faces a different set of pressures: build a useful product, generate revenue, and stay alive long enough to deliver a positive return to his investors.