BlogGuides
AI in bakeries: what the data actually shows about waste

One click, and Google will show my articles more often in your results.
Search "AI bakery waste" and you'll often find reduction percentages as high as 30-34%, published by the companies that sell these systems. The only case reported by an independent journalist, not by a vendor, concerns a handful of Swiss bakeries: after three months of use, waste is described as declining, but no percentage is given at all.
The 30-34% reduction figure has no independent verification
The most-cited number online comes from Foodforecast, a company that sells demand-forecasting software to bakeries and pastry shops. In a testimonial published on the company's own site, Jürgen Kettinger, managing director of Bakery & Confectionery Kettinger GmbH, says: "we used to be around 30%, but now we've been able to reduce this figure to 20% thanks to AI." It's a statement the company chose from among its own clients, with no stated calculation method and no independent study backing it up.
A second figure circulates around PrO4Bake, a European research project coordinated by the University of Hohenheim and funded by EIT Food, whose data collection included Italian bakeries among its participants. According to trade publication EuroShop, the project itself calculated that a 10% cut in bakeries' energy consumption, the target if every bakery adopted its tool, could save the European baking industry up to 760 million euros a year. That's the ideal-scenario projection made by the project proposing it, not a result measured in an actual bakery.
The case reported by an independent source: Swiss bakeries
The only case that comes from a journalist rather than a company selling the tool is the one reported by Swiss public broadcaster SRF on 1 April 2026, then republished without substantial changes by RSI and swissinfo.ch in the following weeks: it's worth saying upfront that these are three outlets from the same broadcasting group running the same piece, not three independent confirmations of each other.
The report describes Go Nina, a start-up founded by economist Matthieu Ochsner and Ferdinand von Hagen, which forecasts every day, product by product, how much to order or produce, combining past sales with weather and holidays. The next day's forecast appears on the till display. According to what the company told the journalist, around 60 businesses use it, mostly bakeries and a few restaurants. The Maier bakery chain, with ten branches across the Swiss cantons of Aargau and Solothurn, has used it since the start of 2026: manager David Giess says that after three months waste is down and revenue is stable or growing, without giving an exact figure. The same report says the Migros and Coop retail chains also use AI systems to plan orders and production.
What's missing, and why it matters to say so
None of the sources, including those favorable to adopting AI, publish a waste-reduction percentage verified by anyone independent of whoever sells or uses the tool. The research behind this article found no documented Italian case of a bakery or pastry shop publishing its own numbers on this. It's a gap that shows up even in the Swiss case above, where the client chain's own manager speaks of waste "declining" without a figure.
Bread waste in Italy is a real problem, even without bakery-level data
There's a documented reason the topic matters: according to the "Il caso Italia 2026" report from the Osservatorio Waste Watcher International, produced by the University of Bologna and Last Minute Market from an Ipsos-Doxa survey and released on 3 February 2026, fresh bread is the third most wasted food in Italian households: 19.6 grams per person per week, after fresh fruit (22.2 g) and fresh vegetables (20.6 g). Total household food waste fell from 617.9 to 554 grams per person per week compared with February 2025, while Italy's entire food chain, fields included, accounts for over 13.5 billion euros of food thrown away each year.
That figure, though, covers what families waste once bread is already in the house, not the unsold stock a bakery throws out at the end of the day: they're two different stages of the same chain, and in Italy no public report measures the second the way this one measures the first.
What it means for a bakery or pastry shop owner
The first investment, before any software, is having sales history in a form a forecasting system can actually read: these tools learn from past till data, so a business that still records sales by hand or on paper starts from zero even after buying the best product on the market. The same principle of keeping a human check on automation applies in a very different case from demand forecasting too: when an AI agent ran a café's ordering on its own in Stockholm, it bought more than four times the pastry the café was able to sell, precisely because nobody checked the orders before they went out.
Faced with a waste-reduction percentage published by whoever sells the system, the useful question isn't whether it's true in the abstract, but how it was calculated and across how many customers, not just the best one. If you're trying to work out whether your business already has the right data for a tool like this to work, before even choosing which tool, get in touch: I'll tell you straight, without selling you something you don't need.