The logic, not the recipe
Twenty-three minutes on BFM 89.9 about where AI earns its place in a commercial kitchen.
In October I went back to BFM 89.9 to join Elaine Boey on Enterprise Explores, in a conversation called When AI Belongs in the Commercial Kitchen. With me was Peter Kua, co-founder of Growthpro.Asia and, with me, of ChefPro.ai, who looks after the technology. Elaine wanted to know where AI makes sense for a food and beverage business, and how an owner can see a return without overspending. These are the points I made on air, with the ones Peter made where they belong.
In a cuisine-driven restaurant, where a meal costs more and is meant to nourish you in body and mind, the biggest headache and the highest cost has always been keeping the right talent. Most of the technology so far has gone into administration. Very little has gone into helping a chef build a menu.
A chef spends a great deal of time on research. We are not born with an encyclopaedia of every ingredient in our heads, and new ones are still being found in our own jungles. I am working in Borneo now, on its biodiversity and its food culture, and that knowledge cannot remain with me alone. If something happens to me, my chefs need to carry on building menus with the same logic.
I want to be precise here, because it is the whole point. ChefPro.ai, the application I co-founded and of which I am the main architect, does not write down my recipes. It holds the logic by which a recipe is made: the framework a cook can use to create the dish.
That is a different thing from writing it all down, and it asks more of a chef. Most chefs do not want to share it, because it is what makes their restaurant different from yours. It is their magic.
It also helps when the brain will not click over. If an ingredient is out of season, or your supplier cannot get it, the system shows you the closest swaps, which supplier has them now, and what they cost. The knowledge no longer sits only with the head chef or the owner, so you can pass authority down with information that has been verified.
This is the biggest pain in the business. Without a system, the owner is held to ransom by the chef. You worry that your chef will leave. What if he is sick? Can the sous chef do it? And if the sous chef is away, who will?
We have been doing this for some time, and it cuts my research time by 80 per cent. It generates the menu and its options, and it can work through very complex permutations: vegan, halal, and so on. It gives me options, and I piece them together to my own preference. The AI is not giving you the recipe. The chef is still creating it; the system gives me sound tools and saves me time.
Take a restaurant that would love to serve wine and has no budget for a sommelier. The system can give the pairing notes and say what is likely to work. I say likely, because the authority stays with the owner, or the person running the kitchen, to accept it or not. It is staged so that you have to agree to each answer before you move on to the next. At no point is the AI doing your work for you. It helps you work better.
When you go to a restaurant, the most important part is service. If the experience at the table is not good, everything falls apart. So imagine taking the same knowledge and passing it to the people who serve the table.
People often say of a cuisine-driven restaurant: James is not here, so the food will not be as good. They have an expectation and a benchmark. With this, we can tell them that he is not here, but everything he knows is.
Peter made the other half of that point. When AI takes away the manual work, people can spend their time on the human touch. At a good steakhouse you pay for the steak, and you pay for hospitality and human connection, which no platform can give you.
As an owner, operator and chef, I could only do creative work after service. That means late at night, when everyone has gone home and it is finally quiet. Imagine 80 per cent of that time given back to you, to put into the parts of the business that make you a better restaurant.
Elaine asked whether this only works at scale. My entire menu for Serumpun Sarawak was created with it: a conservation-driven menu, built on the biodiversity of Borneo, for a new cuisine we showed at Expo 2025 Osaka. But it matters even more to a single restaurant run by its owner with a small team. I no longer need to work on a menu at one or two in the morning. I can do it between the hours of service.
In December the students at Camp of Leaders at Taylor’s Culinary Institute will work through the same application, in a full simulation of a fine dining restaurant.
Peter answered the question every owner should ask first: am I ready? His answer was practical.
- You are ready for AI only once the basics of the business are digital: a point-of-sale system, digital ordering, supplier invoices in digital form, and some kind of inventory system. AI needs data.
- Then look for the work that is manual and repetitive. Checking supplier invoices against delivery orders by hand. Copying daily sales into a spreadsheet to plan next week. Balancing the duty roster across several WhatsApp groups.
- Build a business case before you buy. Ask whether automating that work will justify the investment, in profit, in cost saved or in time. His example was a forecasting system for food preparation that cuts ingredient waste and saves a manager hours each week on inventory; at that point the case makes itself.
- Once it is in place, keep measuring whether the benefit is really arriving.
I put it more simply. If I told you that you could spend more time with your family, instead of seven days a week developing new recipes, what would that be worth to you?
That has always been my motto. If you are miserable, still toiling at remedial work you should not be doing, there is no meaning in it. You are not creating.
He's not here, but everything he knows is here.
James Won, on BFM 89.9
Happy people make happy food.