The mathematics
of flavour pairing

Discover data-backed ingredient combinations and recipe ideas through computational gastronomy.

9k
ingredients
mapped
120
flavour
notes
5m
harmonious
pairings
The Bar LabDiscover flavour trends from the world's best cocktail bars
A painterly image of Rocket
Selected ingredients
Add an ingredient
Search for up to three ingredients to discover their perfect pairings and recipes.
Mathematical methodWe use data science to model flavour profiles, refined by our community of experts
Multi-dimensionalDetailed flavour profiles capture the complexity of ingredients
Unexpected discoveriesFind surprising pairings you won't see elsewhere
Trusted byChefs, bartenders and curious cooks

In Season | November in Britain

Select a seasonal ingredient to see its classic and creative pairings.

Services | Flavour insight

Flavonomics offers a range of flexible services to develop solutions relating to flavour data. We work with restaurants, food brands, and research labs. Partner with us to:

Create and refine menus and recipes
Develop innovative products
Use flavour insights to enhance marketing and branding

Approach | Multi-dimensional flavour analysis

We score every ingredient on 120 flavour notes in 12 families, from citrus and green to roasted and smoky. Each note is tied to the aroma compounds behind it, so a profile shows what an ingredient tastes of and why.

Pairings are scored note by note. We measure which notes appear together across thousands of recipes more often than chance, then rate how well two or three ingredients harmonise and how often cooks actually combine them.

Why Most AI Gets Flavour Wrong

Where chefs would think carefully about the nuances of flavour, most AI will simply string together a sequence of words that seem coherent.

Ask most AI “What pairs with Chenin Blanc?” and it recommends oysters, mussels, and scallops. It's predicting likely answers based on memorised textual patterns, so it offers conventional words associated with white wine.

Flavonomics understands flavour through an internal model of how ingredients actually taste. This allows us to explain why classic combinations work and discover novel pairings that no AI system has ever read about.

Try Deep Flavour

Analysis | Ingredients and flavour

Read our in-depth analysis of more than 4,250 ingredients. Each piece breaks down an ingredient's flavour profile and uses the data to explain what it pairs with, and why.

You can also see what's in season, explore how two flavour notes work together, and browse ingredient trios drawn from the world's top cocktail bars.

Read the analysis