AI Recipe Standardisation: Faster Costing for Chefs

TL;DR: AI recipe standardisation uses tools like ChatGPT to produce accurate yields, portion costs, and method documentation faster than manual spreadsheets. It does not replace kitchen judgement, but it removes the slow, error-prone admin around it.

AI recipe standardisation is not something I ever thought I would be writing about with anything other than mild suspicion. And yet here we are.

I spent the first thirty years of my career writing recipes by hand, costing them on spreadsheets that looked like they had been designed by someone who hated both numbers and human beings, and arguing with suppliers over margins that made absolutely no sense once the food hit the plate. It was slow, it was inaccurate, and it chewed through hours that should have been spent doing actual cooking. So when a junior sous chef suggested I try ChatGPT for recipe costing last spring, I told him to stop being daft and peel the celeriac. Then I tried it at eleven o’clock that night, when no one was watching.

I was genuinely annoyed by how useful it was.

What AI Recipe Standardisation Actually Means

Before we get into the practical side, it is worth being clear about what standardisation means in a kitchen context, because the word gets used loosely. A standardised recipe is not just a written recipe. It is a recipe that specifies exact yields, portion weights, preparation methods, cooking temperatures, and costs per portion. It is the document that allows a kitchen to produce the same dish to the same standard regardless of who is on the section. Anyone who has run a brigade of any size will know this is harder to maintain than it sounds.

AI recipe standardisation, then, is the process of using artificial intelligence tools to help build, format, scale, and cost those documents accurately and consistently. It does not replace the chef. It replaces the tedious administrative labour that most chefs are bad at and deeply resent doing.

ChatGPT for Recipe Costing: What It Can Actually Do

Let me be specific, because vague enthusiasm is useless. ChatGPT can take a recipe written in plain language and help you structure it into a standardised format with consistent units, yield percentages, and portion breakdowns. You paste in your dish, tell it your supplier prices, and ask it to calculate cost per portion. It does this in seconds. Not minutes. Seconds.

It can also scale recipes. If your turbot dish is written for four covers and you need it for forty, you ask. It recalculates. Correctly. It will also flag when scaling creates a technical problem, such as a sauce that will not reduce proportionally at volume, if you prompt it properly. The key word there is ‘properly’. You do need to know what questions to ask, which is where kitchen experience still matters enormously.

What it cannot do is taste. It does not know that your lamb is from a different breed this week and runs leaner. It does not know your head pastry chef always undermeasures butter. These are the things that still require a human standing in a kitchen paying attention.

How to Use Kitchen AI Tools Without Making a Mess of It

The failure mode I see most often is people treating AI as a magic box rather than a very fast assistant. The output is only as good as the input, and in a professional kitchen that means being disciplined about what you feed into it.

Here is a practical workflow that actually functions:

1. Write your recipe in plain language first, with all quantities in grams for solids and millilitres for liquids. Do not skip this step in the hope the AI will guess.
2. Input your current supplier costs. Be honest about this. Use your actual invoice prices, not what you hope to pay.
3. Specify your expected yield. A 1kg piece of beef fillet does not yield 1kg of portioned beef. State the trim loss percentage explicitly.
4. Ask ChatGPT to return the recipe in a standardised format with cost per portion, cost per serving at your target GP, and a suggested menu price at your usual margin.
5. Check the output against your own numbers before it goes into any system. AI makes arithmetic errors occasionally. Not often, but occasionally.

This process, done properly, takes about eight minutes per dish. The old way, done properly, took about forty-five.

Restaurant Recipe Management: Where the Real Value Sits

The deeper value of AI in restaurant recipe management is not in any individual dish. It is in the consistency across a whole menu and across time. When ingredient costs change, as they do constantly and sometimes dramatically, the ability to reprice your entire menu quickly is genuinely important. Doing that manually is the sort of job that tends to either not get done or get done wrong under pressure.

With a properly built AI-assisted system, you update your ingredient costs in one place and ask the tool to recalculate everything. You can then see immediately which dishes have fallen below your target margin and make decisions accordingly. Raise the price. Adjust the portion. Change a component. These are creative and commercial decisions that a chef should be making, not administrative tasks that eat up a Sunday afternoon.

I will say plainly that for smaller independent restaurants operating without dedicated menu costing software, ChatGPT combined with a decent spreadsheet is a remarkably functional alternative to expensive proprietary systems. It will not do everything a purpose-built platform does, but it handles the core tasks well.

Recipe Scaling for Chefs: The Overlooked Problem

Recipe scaling for chefs is one of those tasks that sounds trivial until you get it wrong. Multiply a beurre blanc recipe by twelve and serve it at a banquet and you will understand very quickly why simple multiplication does not always work in cookery. The reduction time changes. The emulsification behaves differently at volume. Seasoning does not scale linearly.

ChatGPT, if you ask it the right questions, will flag these issues. Ask it not just to scale but to ‘identify any steps that May not scale proportionally and explain why’. It is a much better prompt than ‘multiply this by twelve’. The answer is more useful and it forces you to think technically about what you are actually doing.

A Note on Menu Costing Software Versus AI Tools

Dedicated menu costing software such as Apicbase or Netsuite Food and Beverage exists, and for large operations it makes sense. It integrates with purchasing systems, tracks inventory in real time, and generates reports automatically. If you are running multiple sites or a high-volume operation, that infrastructure is worth the investment.

For a single restaurant or a small group, the cost and complexity of those platforms can feel disproportionate to the problem. ChatGPT sits in a different category. It does not integrate with your POS or your stock system out of the box. What it does is think, quickly, about whatever you put in front of it. That is a different kind of useful.

The honest answer is that the right tool depends on the size and complexity of your operation. Many chefs will find that AI tools handle eighty percent of their recipe management needs at a fraction of the cost and time of a full software deployment.

What This Changes and What It Does Not

None of this changes the fact that cooking is a physical, sensory craft that requires skill, attention, and taste. A chef who cannot cook does not become a better cook by using AI. What changes is how much time experienced cooks waste on administrative tasks they are not trained for and do not enjoy.

The sous chef who suggested I try this in the first place has since left to open his own place. He tells me he uses ChatGPT to cost every dish before it goes on his menu. He is turning a profit in his first year. I am not saying the two things are directly connected. But I am not saying they are not, either.