So in the process of building an order page for my app I struggled to come up with a way to suggest other products for a shop to add to its store.
In-built recommendations are one way to do that cheaply with a lot of control, for a task that is not too data heavy.
I took what I learned and made a separate project to show you how it can be done with Thunkable. With this app you can select a type or variety of wine or beer that you like and it will suggest others in a a similar style.
The data was trained based on articles about wine pairings and what experts usually listed as alternatives. Or whether they had very similar tasting notes in general. I have a separate file that I can share for training if you are not comfortable doing it in python.