Recipe Methodology

Every recipe is built on a strict data schema: ingredients with standard units and quantities, numbered steps, times and temperatures, and classifiers (cuisine / category). Research and generation propose drafts from topics with real demand and from real catalog gaps — AI-assisted systems operating under fixed templates, never free-form text without constraints.

1. 01

   Deterministic validation

   Programmatic checks with no room for interpretation: unit and quantity integrity, step ordering, complete fields, and schema conformance.

2. 02

   Culinary rule checks

   Physically possible temperature ranges, coherent ratios and techniques that make sense for the ingredient. Any violation rejects the draft — never an override.

3. 03

   Duplicate check

   Automated matching against published recipes to reduce duplication — no near-identical copies under different names.

4. 04

   Reference-based nutrition

   Nutrition values are computed from a reference composition table where data exists, and clearly labelled as estimates — never invented numbers.

5. 05

   Translation gate

   Translations run through an automated pipeline with fact-lock validation that preserves numbers and units before anything appears in any of the live languages.

6. 06

   Quality gate, then publishing

   Nothing publishes automatically: after passing the checks a recipe reaches a publish queue that requires explicit approval before it goes public.

The platform serves 12 live languages: en, ar, es, pt, fr, de, it, tr, pl, hi, id, ja. Translation is part of the production line rather than a separate manual step, and any language problem can be reported through the contact page with the "translation" category.

Report any error in amounts, steps, times or translation through the contact form with the matching category; reports are handled per the corrections policy.