01
MANUAL CATEGORIZATION
Two methodologies, thousands of products and repeated decisions make the process slow, costly and hard to maintain.
Altera AI
Your protein ratio, without manual categorization.
Altera AI is free. It imports your product files, applies Protein Tracker and WWF separately, flags cases to review, enriches missing nutritional data and generates transparent exports for your internal teams.
The problem
Why measurement is hard in practice.
02
EDGE CASES
Composite products, plant-based alternatives, ready meals, snacks or hybrid recipes: rules must be explicit to stay comparable.
03
INCOMPLETE DATA
Ingredients, protein values, weights or sub-categories may be missing. The system must flag what's reliable, uncertain or to be corrected.
Workflow
From raw file to auditable reporting.
01
Create a project
Define the scope and choose the methodology: Protein Tracker, WWF or both.
02
Import the file
Upload a retail CSV and map product, category, brand, ingredient and nutritional columns.
03
Classify
Combine deterministic rules and AI classification to produce a first categorization.
04
Review
Surface uncertain products, edge cases and required corrections in a validation table.
05
Enrich
Complete missing nutritional values where possible, drawing on the open NEVO nutritional database.
06
Export
Generate ratios, methodology reports and categorized files ready for your internal teams.
Methodologies
Two methodologies, never mixed.
Protein Tracker
Measures the mix of animal, plant, composite and out-of-scope protein, to track the commercial evolution of the animal / plant ratio.
WWF
Classifies products by food groups and sub-groups, with particular attention to composite products and assortment trajectories.
A single project can hold both classifications, but Altera AI never creates a hybrid methodology.
Human control
The AI suggests. The methodology decides.
Altera AI doesn't silently accept ambiguous cases. Uncertain classifications are flagged for review, corrections are kept, and exports document sources, confidence levels and rationales.
- Uncertain cases flagged
- Corrections kept
- Visible rationales
- Auditable calculations
Security
Sensitive data kept separate from the AI.
Sensitive commercial data is never sent to the classification model. Volumes, sales, margins, costs, store data, suppliers and strategic information stay separate. The AI works only on the descriptive information needed for product categorization.
Exports
Outputs your teams can use.
01
Protein Tracker report
02
WWF report
03
Categorized Excel
04
List of cases to review
05
Sources & confidence levels
06
Internal steering base
Contact
Ready to test Altera AI on your product data?
A first call lets us assess your scope, the available data and the best method to produce a usable baseline.