A simple intelligence layer can translate mood and taste preferences into a small number of relevant drink choices—then explain why each recommendation fits.
The recommender can ask about mood, coffee intensity, sweetness, milk preference, flavor family, temperature and caffeine preference, then suggest suitable menu options with a clear reason.
An optional profile can remember preferred sweetness, strength, milk, flavor and temperature.
Recommendations should show the matching logic instead of presenting a mysterious score.
Preference data should be minimal, transparent and optional; a guest should not need to provide personal data just to get a coffee suggestion.