Reducing food waste with artificial intelligence - how does it work?

Dear Blog Readers, 

Food Waste and Artificial Intelligence (AI) - two terms you don't necessarily associate with each other! And yet they belong together, because we have increasingly large amounts of data at our disposal, which we can use intelligently for more sustainability in the food industry.  

We wanted to find out in person from our CEO and founder, Justus Lauten, how exactly artificial intelligence leads to less food waste. In the interview, he explains how it works.

We use our software based on artificial intelligence to reduce food waste in the food industry. How does that work exactly?

Justus: 

Our self-developed artificial intelligence determines exact sales forecasts for food production companies. Based on this, our customers receive an order proposal every day that is tailored to their needs. This specifies the exact quantity of items to be ordered so that only what is actually sold at the end of the day is produced. This helps our customers to significantly reduce their food waste. But not only is there less food waste, fewer resources are wasted producing and selling the food that is thrown away. 

How does the AI find out what sales volumes are to be expected?

Justus:

The optimal sales volume results from various factors. We connect our AI software to the customers' merchandise management or POS system and thus obtain the customers' historical and current data. This is, for example, sales and order data. We combine this with relevant, external factors such as weather data, public holidays and vacation periods. From this, the algorithm independently recognizes patterns and dependencies and determines the forecasts. The special feature: the AI learns independently and recognizes trends from the first data point - so it doesn't just replicate the past.

Are there products where this forecast works better than others?

Justus: 

In any case, it is easier to make predictions for items where a lot of data is already available - preferably also over a longer period of time. The more data available, the more accurate the AI results. For example, in the bakery, predictions for bread and rolls work particularly well because the products have been sold daily for many years and therefore a lot of historical data is available to us. Challenges arise when items are newly added to the range. Then we don't yet have any internal sales or order data with which to train the AI. Nevertheless, our AI can already derive the forecasts from the first data point.  

By what percentage can artificial intelligence reduce food waste in practice?

Justus:

On average, we reduce our customers' food waste by 30% with our software. This corresponds to an annual saving of around 2000 tons of food waste and a CO2 equivalent of 9500 tons. To make these numbers a little more tangible, that's the equivalent of about 300 truckloads of food waste or as much CO2 emissions as 3000 flights to Australia that we save per year with our software. This motivates us every day on our way to our big goal of fighting global food waste. The potential of our software is huge, because every year 1/3 of food is wasted in food retail and catering worldwide.  

What is the benefit from our customers' point of view - in addition to more sustainable food handling?

Justus: 

In addition to the sustainability aspect - i.e. minimizing food and resource waste - the most important benefits for our customers are process automation and optimal availability of all items at any time of day. Through automation, our AI relieves the employees, who no longer lose time in production and order planning and can concentrate on their core tasks. This leads to more effective staff deployment and fewer errors. In addition, AI contributes to optimal availability of goods. In this way, we help our customers not only to reduce their food waste, but also to avoid selling out of individual items too early and thus to increase their sales. In a free pilot phase, customers can test our solution and we show them the added value on the basis of detailed analyses. In this way, they see in black and white that the use of artificial intelligence is worthwhile both ecologically and economically.

Is the use of AI a big change for our customers?

Justus: 

Using our AI software is not a big change for our customers, because foodforecast does the work and we plug into their existing processes. We build the interface to the enterprise resource planning system to pull the data we need to train the AI. We then transmit the forecasts to the customer, who can view them directly in their merchandise management or POS system, or in an external app. One potential challenge lies in the quality of the data, as this plays an important role. That is, how accurately the documentation of the data was done in the past. If orders or sales have not been entered correctly, there can be discrepancies between human planning and AI forecasts, and skepticism sometimes arises. That's why we also use our pilot phase to put possible concerns aside in the process.

It all sounds so easy, doesn't anything ever go wrong?

Justus:

Of course, not everything always goes according to plan for a startup. When I was looking for customers in the beginning, for example, I approached the organic baker around the corner from me to see if he would be interested in our AI forecasts. Without further ado, we agreed and I plugged a USB stick into his cash register to get the necessary data from the bakery to train our AI. I waited 5 min, 10 min, 15 min and the transfer was still not completed. In the meantime, a long line of customers had already formed. The salespeople had to write down all the orders by hand and were getting restless because they didn't know all the prices for the baked goods by heart. After half an hour, all employees were finally able to access the cash register again. After that, we built a more professional interface to our customers' POS systems.