AI recipe reformulation: faster reformulation with fewer trials
A supplier announces they’ll stop manufacturing ingredient A from July. Or a regulator bans additive E. Or you want to cut costs by substituting an expensive component. Traditional reformulation takes 4–12 weeks of laboratory trials.
AI reduces this to days.
Where AI reformulation helps
Ingredient substitution
Supplier changed or discontinued production → AI proposes alternative ingredients with similar physical and chemical properties and suggests an adjusted ratio in the recipe.
Regulatory change
EU bans an additive → AI identifies permitted alternatives with a similar function and proposes a new formulation.
Cost optimization
Price of ingredient X rose by 40 % → AI proposes a substitute or reduced proportion without impacting the final product properties.
New market / nutritional requirements
You want a product with less sugar, allergen-free, or higher protein content → AI proposes a reformulation targeting those nutritional goals.
How AI works with recipes
Ingredient database
The system contains a database of ingredients with physical, chemical, and functional properties. For food: flavor profile, viscosity, pH, gelation temperature, allergenicity.
Dependency model
Based on the history of recipes and laboratory test results, AI learns dependencies: “increasing ingredient X by 5 % at processing temperature T raises viscosity by 12 %.”
Candidate generation
For the given constraints (cost, availability, regulatory compliance, target properties) AI generates 3–10 candidate recipes ranked by probability of success.
For food industry: nutritional declaration automatically
For every proposed recipe, AI automatically calculates:
- Nutritional values per 100 g and per serving
- Allergen declaration
- Ingredient list in descending order of weight
- Additive codes (E-numbers) and their regulatory status in the EU
