Control food cost and recipe costing with AI in a restaurant

Food cost is the number that decides whether your restaurant makes money or just moves it around. Most kitchens calculate it once, write it down in a spreadsheet and never look at it again until the margin has already evaporated. That is the problem. Purchase prices change every week, waste piles up without anyone recording it, and the menu's recipe costing goes stale within days.

Artificial intelligence solves this without a data team or an expensive ERP. A well-built system calculates the real cost of every dish, updates it on its own when ingredient prices rise, and alerts you the moment a dish stops being profitable. In this article you'll see how it works, what results to expect, and how to get started in a matter of days.

What food cost is and why it slips out of control

Food cost is the percentage that raw ingredients represent against a dish's selling price. The healthy range sits between 28% and 35%. A 12-euro set lunch should stay around 25-28%. An à la carte menu with a 35-euro average ticket holds up well at 32-35%. Fine dining can drop to 20-25%, and a seafood restaurant can reach 40% and still be profitable if the ticket allows it.

The problem is not calculating it once. The problem is keeping it there. When food cost stays above 35%, there's almost always one of three causes: purchase prices that rose without anyone adjusting the menu, waste and shrinkage nobody measures, or a recipe costing that was wrong from the start. All three are invisible in a static spreadsheet. All three are obvious to a system that cross-references purchases, sales and stock in real time.

Waste is the quietest part. In 2024, Spanish hospitality threw away 28.03 million kilos and litres of food, though it was the sector that cut its waste the most, down 8.8% on the previous year. Globally, wasted food in the food-service sector exceeds 100 billion dollars a year. Every kilo that goes in the bin is margin you've already paid for and won't get back.

What AI does with your recipe costing

The idea is simple: connect what you buy, what you sell and what you have in the storeroom, and let the system run the calculations you don't have time to do.

Automatic, always-updated recipe costing

Instead of redoing the recipe costing every time a supplier changes, the system reads your purchase invoices, extracts the price of each ingredient and recalculates the cost of every dish instantly. If the price of oil goes up 15%, you know immediately which dishes have become less profitable and how much you'd need to raise the price to keep the margin.

Margin per dish in real time

Not every dish performs the same. AI cross-references the real cost with POS sales and tells you which dishes are stars (high margin, high volume), which are dead weight (low margin, low volume) and which are worth redesigning. This turns the menu into a profitability tool, not a wish list.

Deviation and shrinkage alerts

The system compares what you should have consumed based on sales with what actually left the storeroom. When the gap spikes, an alert fires. That's how you catch shrinkage, portioning errors and even theft before they eat into the month's margin. Waste-tracking programmes cut raw ingredient cost by 2% to 6%.

Purchase and production forecasting

By cross-referencing sales history, day of the week, events and even the weather forecast, AI anticipates how much you'll sell and how much you should produce. Forecasting systems reduce overproduction by 15% to 25%, and computer-vision waste-control tools have been shown to cut it by up to 30% within three months of deployment.

A concrete example

Picture an à la carte restaurant with 40 dishes and a 30-euro average ticket. Before, the recipe costing was reviewed quarterly, by hand, whenever there was a gap. In practice, that meant working with costs from three months ago.

With an AI system connected to invoices and the POS, that same restaurant detects that three fish dishes have risen enough to jump from 33% to 41% food cost. The system flags it the same day. The fix is immediate: adjust the portion, renegotiate with the supplier or raise the price by 1.50 euros. Multiply that correction across dozens of dishes and several months, and the impact on the bottom line is direct.

How to start fast and cheap

You don't need to change systems or stop the kitchen. Here are the steps:

Start by connecting what you already have. Your POS records sales and your suppliers issue invoices. That's enough to get going. AI reads the invoices, even when they arrive as a PDF or a photo, and turns them into cost data.

Digitise the recipe costing of your best-selling dishes first. You don't need to load the whole menu on day one. The 20% of dishes that generate 80% of sales already give you most of the value.

Turn on deviation alerts. Set a food cost threshold per dish and let the system warn you when it's crossed. That's what turns control into something continuous rather than a quarterly exercise.

Review the data once a week. Fifteen minutes to see which dishes have deviated and decide whether to adjust portion, price or supplier. That's all the maintenance it takes.

The ROI: why it pays off

The maths is simple. If your restaurant bills 400,000 euros a year at a 34% food cost, you're spending 136,000 euros on raw ingredients. Cutting that food cost by just 3 points, to 31%, means 12,000 euros a year that stay in your pocket. And 3 points is a conservative target when you move from quarterly control to automatic, daily control.

Then there's time. The hours a head chef spends redoing recipe costing and reconciling spreadsheets become hours in the kitchen or on the floor. Spain's new food waste prevention law pushes in the same direction: measuring and reducing waste is no longer optional, and a system that does it on its own saves you the work of complying.

Start today

Controlling food cost with AI is not a months-long project or a thousands-of-euros investment. It's connecting what you already have, starting with your key dishes and letting the system run the numbers you don't have time for today. The difference between a restaurant that survives and one that makes money is often those 3 points of food cost.

At Obsidy we build this kind of automation fast and at a cost that pays off from the first month. If you want to control your restaurant's food cost without setting up a data department, write to us at hola@obsidy.com or visit obsidy.com and we'll build it with you.