It's 6:30 a.m. on a Monday morning in Oslo. The canteen manager at Hausmanns Hus opens his email and is told that 347 guests will be coming for lunch today. Tomorrow it will be 289. Next Monday 412. Four weeks into the future, the numbers are clear, broken down day by day. The precision? 95%.

Welcome to the new reality of the canteen.

From gut feeling to millimeter precision: How prediction cuts food waste and costs in the canteen

Ole Aabel at Izy AS has heard the reactions many times. “People don’t understand that it’s possible. They think we’re joking when we say we can tell them exactly how many lunches they’re going to sell four weeks in advance.”

It's understandable. In an industry where even experienced chefs base purchasing decisions on gut feelings and historical experience, this precision comes as a surprise, as decisions are supported by actual sales data, weather and calendar. But there's a rational explanation: massive amounts of data, machine learning and a deep understanding of how Norwegian office buildings work.

Machine learning on the kitchen counter: How to make the forecasts

What's really going on behind the scenes? Saurav Pandey, the backend developer who built much of the system, explains it this way: “In theory, it's about using historical data to make forecasts. But in practice? It's about the cafeteria manager not having to lie awake at night wondering if she ordered too much cod.”

Behind each forecast lies a complex puzzle. The system gobbles up transaction data from the Izy app and mAIfood, cross-checking with weather forecasts, holidays and traffic conditions. Is the building planning a big event? It's registered. Is it going to snow? The system knows that fewer people will make the trip to the office.

All of this is collected, weighed, and analyzed. The result is a number you can trust. Nine times out of ten, it hits the nail on the head.
“Experienced chefs are good,” admits Pandey. “But they can’t compete with this. Our brains simply don’t remember what happened on the same Tuesday last year, or how the weather affected sales the last time it snowed. The machine does that.”

Ole Aabel hastens to add: “But they shouldn’t compete with this either. It’s not a competition. It’s help in everyday life. Technology that makes things flow better, that makes the job easier. The chef can focus on what he does best, making good food, and stop being a fortune teller.”

The system's three dashboards: Sales, CO 2and staffing

The system consists of three dashboards. Each solves its own problem. Together they create something bigger.

The sales dashboard is the heart of the operation. Here, the canteen manager gets the answer to the eternal question: How much food do we need? The forecasts for the current week drive the kitchen production. The figures for next week go straight to the purchasing department. No surprises. No panic orders at seven in the morning.

The CO₂ dashboard is the future of the game. Everything that can be measured is measured here: food waste, energy consumption, emissions from raw materials. The system doesn't just give numbers, it suggests concrete actions. Replace beef with chicken on Tuesdays. Lower the temperature in the refrigerated counter by one degree. Small changes that have big effects.

The resource dashboard calculates staffing needs. This is the most controversial. “When we talk about optimizing staffing, it creates insecurity,” admits Ole Aabel. “People hear ‘efficiency’ and think ‘downsizing.’ But it’s not about throwing people away. It’s about using them smarter.”

Resistance to change. It is perhaps the biggest obstacle of all.

Proven success: The solution is used in a growing number of canteens

Compass Group has 22 canteens that actively use Izy Prediction in their daily operations. That's just the beginning. Several major players are now signing contracts, and with new customers coming in later this fall, Izy expects to at least double the number by the end of the year.

Every Monday morning, an email lands in the chefs' inbox. “Here are the numbers for the week. And a little peek at next week too.” Ole Aabel smiles. “We could have given them four weeks, but they obviously can't handle more than two. In the dashboard, there's two plus two weeks. A starting pad.

The analysis department has access to everything and can dive as deep as they want.” Why does this work so well? Saurav Pandey points out something essential: “We operate in a very niche. Norwegian office buildings. They behave quite similarly.” First day of winter in Norway, then fewer people come to work. This applies to almost all buildings. The patterns repeat themselves. 

Therefore, the system needs surprisingly little data. Just one month of transactions is enough to start making solid forecasts. Not years. One month.

Concrete savings: Cut 30% on purchasing and 15% on staffing

Ole Aabel leans forward when the conversation turns to finances. “Staffing: typically 10-15% savings. Purchasing: up to 30%, depending on starting level and menu flexibility. And food waste? It plummets.”

Let's take a step back. What really happens in a canteen kitchen today? The chef is there at six in the morning and has to make decisions. “What if more people come than I think? Better to cook too much.” We all know the result: Overproduction. Food waste. Lost money.

“Food waste is not just what you and I throw in the canteen,” Ole clarifies. “It’s production waste. Raw materials that get too old in the fridge. Preparations that are never used. When you know exactly how many people are coming, you can produce precisely.”

Imagine the difference: You know there are 347 guests coming today. Not 300. Not 400. 347. You make 350 servings. Three are left over. That's waste you can live with.

When sustainability becomes a competitive advantage

The CO₂ dashboard measures everything. Food waste. Energy use. Emissions from raw materials. It doesn't just give numbers, it points to solutions.
“We believe contracts will be won on documented sustainability,” says Ole Aabel.

His voice becomes more serious. “We know that companies have lost contracts because they couldn’t document well enough.”
The EU is tightening up. New reporting requirements are rolling in. Large companies must document their carbon footprint. Not approximate. Precise.

“The product we are building now,” explains Ole, “is a CO₂ product that tells you exactly where you stand.”

For farm owners, this opens up completely new doors. Measures that have been proven to reduce food waste can be classified as sustainable according to the EU taxonomy. This means potentially more favorable financing. Green loans. Lower interest rates.

In concrete terms, this means that measures that reduce food waste can count towards the EU’s taxonomy under the circular economy objectives. When you can document real reductions in food waste, not just claim it but prove it with data, it strengthens the case for green loans. Banks and investors are looking for exactly this kind of documentation.

Sustainability is no longer just a nice thing to have. There is money on the table.

Expanded potential: From canteen to optimization of the entire building

Ole Aabel sees something different when he closes his eyes. He envisions big changes, a completely new way of doing FM business.

“Imagine a pool of employees. They get a message the day before: Tomorrow you need to go to building A. Sales are increasing there. They need you.”
He continues. Large FM players centralize. For example, companies can collect all production of baked goods in one place. Deliver to all canteens based on the forecasts.

“You remove 50% of the operating cost. Maybe more.”

But. And here comes the big but.
“This requires changing business models. It’s the hardest thing there is. The resistance is enormous.”

Change is scary. It's not the technology that's the problem. It's the fear of the unknown.

Trust in AI? How to overcome the fear of the unknown

A major customer said no. For free testing. Zero kroner. No obligations. No thanks.

“Why?” Ole Aabel asks the question rhetorically. “It creates insecurity. It touches on many roles in the organization.”

Here lies the core of the entire AI problem. Not the technology. The trust.

“It’s no problem to explain what we do,” says Ole. “Everyone understands what we do. But how? They don’t understand. And when you don’t understand how, trust is lost.”

But then there are the others. Those canteens that said yes. Compass Group took the chance.
Every Monday morning the system is proven anew. 347 guests today. The result at lunchtime? 343. The deviation? 1%. The next day the same story. And the day after that.

“We have 95% certainty. We miss by 5%,” says Ole. “But people don’t believe it until they see it with their own eyes.”

Now they see it. The purchasing budget is down 30%. Personnel costs are falling. Food waste is plummeting. The numbers don't lie.
And suddenly? More people are knocking on the door. More people want to test. More people want to see the evidence for themselves.

“Trust is not built with PowerPoint,” Ole smiles. “It is built with evidence. And now we have canteens full of evidence.”
The story of the no? It's becoming increasingly rare. The story of the yes? It's spreading.

Izy's unique advantage: Niche data from Norwegian office buildings

There are competitors. Prediction tools too. But Izy has something unique.

“We’re sitting on niche data that probably no one else has,” explains Saurav Pandey. “Izy as a platform has been around for five or six years. Millions of transactions from Norwegian office buildings. That data is gold.”
It's not random. It's strategic. Everything in the Izy ecosystem talks to each other. 

The data flows automatically into the forecasts. The Izy app? Same thing. No manual transfers. No Excel sheets to match. Everything in one place.
“Everything in one place.” It’s not just a slogan. It’s the architecture. The philosophy. The competitive advantage.

While others deliver fragmented solutions that need to be stitched together, Izy has built a holistic system where each part reinforces the others. When mAIfood does its job better, forecasts become more accurate. When forecasts become better, purchasing becomes smarter. The spiral goes upwards.

The dream scenario

Who is the ultimate customer?

Ole Aabel thinks for a moment. “Right now? Hospital. Think about it. You know when patients are coming. You know the staffing. Enormous amounts of data flow through the systems every day. The need for efficient resource management is massive. “That’s where we want to create huge value.”

High schools are also knocking on the door now. But it's getting tougher there. "We don't know what makes students come to school or not. It's getting much harder to predict." Too variable. Too unpredictable.

But hospitals? Normal cafeteria operations? They have what they need there: Patterns. Data. Predictability.

The entire building in real time: The optimization of the future

But canteens are just the beginning. Ole Aabel sees much further.

“Now we are concentrating on food, resources and purchasing. But potential? Utilization rate in buildings. Meeting rooms that are empty. Power consumption based on actual activity.”

The picture he paints: An entire building optimized on real-time data. The system knows how many people will arrive tomorrow. The lights on unused floors are turned down. The temperature is adjusted. Meeting rooms are automatically allocated based on need.

“That’s the future I see,” he says. “A system that optimizes the entire building, not just the cafeteria.”

Data becomes intelligence. Intelligence becomes action. Action becomes savings. And environmental gains. And a better working environment.
All based on the system knowing one thing: How many people will come tomorrow?

Data security and GDPR: Security in data management

But with large amounts of data also come large obligations. Privacy. GDPR. Compliance.

“We take data security seriously,” emphasizes Ole. “The system is built with privacy as a foundation, not as an afterthought.”

Izy complies with all current EU data handling requirements. Especially relevant now that new reporting requirements are rolling in. Companies need to be able to document not only what they do, but how they do it, safely and legally.

Read more about the new EU reporting requirements in canteens

When the system is expanded to optimize entire buildings, this becomes even more important. But the principle remains: Data should create value, not worry.

Back to the kitchen counter

The end of the story goes back to the beginning. To the canteen manager at Hausmanns Hus who opens his email on Monday morning. 347 guests today. 289 tomorrow. The numbers are there, clear and distinct.

A year ago she would have had to guess. Order some extra salmon just to be safe. Hope for the best. Now she knows. And that knowledge changes everything.

This isn't about AI, not really. It's about the everyday life of a canteen manager who has to decide how much salmon to buy on Friday. About the farm owner who has to document sustainability to win contracts. About the FM manager who has to optimize resource use without losing quality.

“We’ve said it from day one,” says Ole Aabel. “It’s not the solution itself that creates value. It’s the data that comes out of it. This is the first time we’ve really put our own data to use to create value for our customers.”

And maybe that's exactly why this works. It's not technology for technology's sake. It's technology in the service of people. For chefs. For buyers. For everyone who runs canteens and buildings.

The system learns. It improves. Every day, in every single canteen – from Hausmanns Hus in Oslo to canteens all over the country – value is created. Less waste. Lower costs. Greener operations.
Gut feeling is still allowed. But now it has an ally. Data that points in the same direction. Precision that confirms instinct.

Or correct it when needed.

Ready to take the leap from guesswork to precision? Visit izy.no/izy-prediction or contact our team to hear how we can help your business.

Ready for change? Here's how to get started in 4-6 weeks

The path from interest to operation is shorter than you think. Three steps:

• Step 1: Book a demo with us (week 1)
  Contact us for a no-obligation demonstration of our solutions. Together we will map out what data is available and what is needed for your canteen or building. Transaction history from the cash register or existing systems. We only need one month of data to get started. We will figure this out together. Then we will move on to step 2, if you wish.
• Step 2: 30-Day Model Building (Month 1)
  The system learns the patterns in your cafeteria. No installation of new hardware. No downtime. The data flows in, the model is trained, the forecasts become increasingly accurate.
• Step 3: Weekly operation (from month 2)
  Every Monday morning: the numbers in your inbox. This week's forecasts plus a look at next week. The dashboard shows deeper analysis when you need it. The analytics department can dive into trends.

Total time from yes to first prognosis: 4-6 weeks.

Izy Prediction in numbers:

• 95% accuracy on forecasts four weeks ahead
• Up to 30% savings on purchases
• 15% reduction in personnel costs
• 20+ canteens actively use it every day (expected to double by the end of the year)
• 1 month of data is enough to start
• 3 dashboards: Sales, CO₂, and resources