15 recipes. One shopping list, calculated automatically.

The context

Wankas is the brand founded by Milena Pecho, winner of the 2024 Cuillère d’Or.

After five years in financial auditing, she decides to follow her passion and starts an event catering business.

It all starts with the recipe specification sheet: the recipe for a dish, with the exact weight of each ingredient. The catering collective she belongs to requires one for every dish. It serves three purposes:

• calculating shopping quantities. For 200 portions, the sheet tells you how much of each ingredient to buy.
• reproducing the same flavour every time.
• helping a kitchen assistant work independently. They follow the document without needing her.

The problem? Filling them in takes time. Everything happens in spreadsheets, one recipe at a time. The other caterers do it, but the task remains tedious, especially as the collective’s template does not add up ingredients across different recipes.

Milena built a file that does. It turns her orders into a shopping list and a production schedule. But the business is growing, and the file has reached its limit.

Several catering jobs, one production day

She can prepare food up to 3 days before an event. A single production day can therefore cover several catering jobs, which she groups together manually before shopping.

One ingredient, 15 recipes

The same ingredient can appear in 15 dishes made during a single production session. Adding up quantities and preparation stages recipe by recipe takes time and can lead to errors.

One forgotten ingredient, half a day lost

A missing ingredient in the middle of production means a round trip to the wholesaler. Production stops.

Kitchen assistants who depend on her

The printed schedule is not enough. Kitchen assistants keep coming back to ask how to prepare the food and in what quantities.

The solution

A custom SaaS application. Upload the purchase order and the catering job is created automatically. Previously, every order had to be entered manually.

For each production day, the tool brings together all catering jobs and calculates a shopping list by ingredient.

It then generates a schedule organised by kitchen stage: cutting, preparation, cooking, assembly and plating. Quantities are already calculated.

The schedule can be viewed by dish or by ingredient. Take onions that need slicing for 3 different dishes. The ingredient view shows the total, so they can all be sliced at once. Kitchen work is organised by stage; the schedule follows the same logic.

The application is responsive and works on mobile. The schedule stays readable on a phone during production.

Stack: Claude, Lovable and Supabase. First version delivered 4 days after the interview.

Method: BMAD, a framework that lets AI agents work as a product team. Each agent has a defined role. Every stage requires my approval.