We waste 1.6 billion tons of food every year while 25 million starve and another billion are malnourished. Can AI fix it? If one startup in Berlin is successful, just maybe. The global food supply chain is mind-bogglingly complex. Tens of millions of farms feed millions of grocery stores and restaurants, which in turn supply almost eight billion people their daily food. Plus of course there are transport companies, wholesalers, distributors, processors, and delivery companies. Put it all together, and you have a massive web of producers and consumers joined by everyone in between that literally encompasses every single human being on the planet. And all of them, of course, dealing with perishable product. That complexity and chaos is what startup company SPRK.global, is trying to fix. The company is one of eight winners out of 2,400 entrants in “the world’s largest startup competition for entrepreneurs addressing global challenges,” the Extreme Tech Challenge. The goal: use AI to understand the flow of food and reduce waste. That should lead to both less hunger and less over-production.
“Half of the food that gets produced gets wasted sooner or later,” CEO Alexander Piutti told me recently on the TechFirst podcast. “Once you move into understanding patterns — why there are food waste cases — you understand these patterns and see they come in a regular fashion … we can move from reactive to proactive, to anticipating, to predicting with a certain probability.” Food waste is an ecological problem as well as a humanitarian one. Overproduction uses resources like fuel, water, fertilizer, increasing greenhouse gas emissions in the process. And then when food is wasted, you need landfill space and there’s even more CO2 production.
“Food waste is one of the largest CO2 emitters,” Piutti says.
The question is: can AI solve it?
But before AI kicks in, the system needs to understand supply chains and economics. For instance, if there’s oversupply in one area, that doesn’t mean that stores or distributors with too much want to give it away to competitors with too little. On the other hand, they’re much more likely to give it to food banks and other NGOs that can provide it to those who need it and would not have been able to afford it. So SPRK needs rules like this in order to function in the real world of actual economic conflicts.
“Once we have these rules, we can inject them into the technology,” Piutti says. “The technology takes over … and matching between oversupply and demand … becomes more intelligent over time.” In a way it’s similar to Google’s Project Loon, where self-governed internet balloons deliver cellular connectivity and web access over rural Kenya. Where they float in the wind is governed different winds at different altitudes, and the machine learning algorithms that deploy the balloons has learned over time — and is still learning — how to navigate in order to to maximize coverage of the required territory. One place to start is food banks. They typically operate with a phone and maybe a spreadsheet, but the software-as-a-service onslaught that’s taken over so many industries with, essentially, business operating software has not touched food banks or similar NGOs. So SPRK is building software that they can use to manage their own operations as well as collaborate: one has too much, the other has too little. Sharing now will be reciprocated in the future.
In addition, Piutti says, the software will give them better ways to access food at lower prices. “They purchase food in a very normal fashion, they don’t get discounts,” he told me. “If we can connect the dots conceptually and say like, well, what if we distributed this food oversupply to the folks in need … they become a volume partner.” In other words, most NGOs that distribute food buy it on the open market. If distributors have too much, SPRK can access it — under conditions that it won’t re-enter the commercial market — and provide it to food banks and others at much lower prices. That saves the NGOs about 50%, it reduces food waste, and it provides operating revenue for SPRK. “So win, win, win situation,” says Piutti, who worked for Yahoo back in the day and advises multiple other startups. “This is how we think about the solution, you know, without generating channel conflicts, to be transparent to our supply partners, because they need to approve.” The goal isn’t to reinvent the wheel: just to use what’s there a little more intelligently.
While SPRK is starting in Berlin, the company has global aspirations from the U.S. to Asia, in what could ultimately be a massive cloud-based platform for the equitable, efficient, and effective distribution of food. At least, oversupplied food. “It’s like one big platform, call it ‘Amazon of Food Oversupply’ if you like,” Piutti says. It’s a big vision and there’s a lot to build before it’s reality. Over time, if SPRK is successful, it will likely extend beyond distributors and NGOs to producers and consumers, injecting intelligence and prediction into farming and prediction, and distribution and shipping. It’s a big job, and it’s not just about how various agents and businesses in the industry interact with each other. It’s also about how businesses operate internally, where there’s also a lot of waste. SPRK’s software could help them too, Piutti says.
“If you are … a large food retailer and you have hundreds of supermarkets that you ship to, even those still have the means to understand who’s oversupplied, who’s undersupplied.” This isn’t his first startup focusing on global food. A previous startup called Sharecy, still visible on Piutti’s LinkedIn profile, had a similarly massive goal: “Our vision is a world without food waste where everyone — including future generations — have enough to eat and thrive.” SPRK is now his second kick at the can. Business results aside, if no-one starves, and we stop letting half of global food production rot, Piutti will know he’s achieved his goal.