AI-powered technologies are helping food businesses improve forecasting to reduce food waste and save money, while also transforming deep ocean fish farming to create more sustainable, reliable food production.
In the United States, 40% of food waste comes from restaurants, grocery stores and food service companies, according to Recycle Track Systems, a sustainable waste collection services provider. Effective supply chain forecasting plays a pivotal role in reducing food waste across the supply chain. Yet, most businesses aren’t accurately predicting almost half the time.
“The average accuracy of demand forecasting is around 60%, which means that most businesses are getting it wrong 40% of the time,” said Drew Ryder, strategic innovation officer, Vizen Analytics.
“In the food industry, that's really important, particularly on perishables because they have a shelf life. If I pre-purchase something that has a shelf life of two weeks, and I can't move it in two weeks, I'm either heavily discounting or I'm literally throwing it away.”
Traditionally, companies have relied on deterministic software, which uses a formula to predict an outcome based on unique variables. Deterministic software will create the same output every time the same input is provided, which means it’s essentially static.
“That's how most businesses still are running their businesses, with deterministic software, because it's reliable, it makes sense,” Ryder said. “But it's, quite frankly, not.” While grocery retailers may be able to forecast their needs for three weeks or a month using similar software, forecasting beyond three months becomes difficult, he explained.
With the help of AI, Vizen Analytics’ forecasting platform enables companies to not only forecast their near-term needs more accurately, but makes it possible for companies to accurately forecast their purchasing needs for individual SKUs as much as a year in advance.
“The magic of AI is that if we were to build some initial models for a particular process, they're not static, right?” Ryder explained. “It's not like we build a model, we hand it to the client and say, ‘Hey, here you go. Just use this formula constantly.’ The model is constantly ingesting new data into it. Initially, it's just the customer data [and] historic data, but our differentiator is we are bringing external data into the models, as well.”
Local events, weather, and socioeconomic and political factors are examples of external factors that can affect consumer shopping behaviors and, thus, Vizen Analytics’ forecasting models.
“So, we may have a forecast today that says ‘X,’ and then wait five days and it might have changed because some factor, using what we call ‘causal modeling,’ will impact that result,” Ryder said.
In a pilot of a large retailer using publicly available data that comprised about 10 stores and $100 million in inventory over a two- or three-year period, Vizen Analytics’ modeling software improved forecasting accuracy by 20% to 25%—resulting in savings of $1 million in overhead costs in one year. Additional savings may also be generated from more accurate purchasing, such as reduced warehouse and labor costs associated with less inventory.
“What we’re doing—this is really the power of AI,” Ryder said.
AI is also making waves in aquaculture, where it’s created a path for Forever Oceans to sustainably raise fish offshore in deep water.
Historically, challenges of offshore fish farming have forced operations to remain in near-shore locations. For example, offshore fish farming operations are subjected to the open ocean environment, which means enclosures must be able to withstand strong currents, wind and waves, according to Bill Bien, CEO of Forever Oceans.
Farm fishing also brings with it environmental woes that have come under fire in recent years, including pollution of the surrounding waters caused by fecal matter and uneaten food that results from overfeeding.
With the help of AI and enclosures that can withstand hurricane force winds, Forever Oceans has overcome such challenges.
“A big part of our approach is remote wireless control and monitoring, which is important because historically the industry has been constrained to near-shore operations due to logistical limitations,” Bien said. “Our more automated, satellite-controlled approach addresses this historic challenge. And our monitoring technology interacts with our AI-driven software to determine when fish ought to be fed and harvested. This approach limits excess feeding and optimizes feed conversion ratios, further reducing our environmental impact.”
The company aims to be the first to master deep-water offshore aquaculture at scale—where “AI plays a central role in our approach,” Bien said.