Nutrition transparency
MealThinker is designed to make generated meal plans fit your targets more consistently. This page explains what the numbers mean, how they stay aligned across the app, and where estimation still has limits.
MealThinker estimates energy needs with the Mifflin–St Jeor equation, then applies the activity level and goal you provide. Protein, carbohydrate, and fat targets are derived from that calorie target and your selected diet. You can review and change these inputs in your profile.
For supported planning flows, MealThinker first builds a structured meal from measured ingredient components. Its calories and macros are locked at that stage. The recipe-writing step expands those same components into ingredients and instructions instead of inventing a second set of nutrition numbers.
Ingredient quantities and nutrition values scale together when servings change. Final quality checks reject malformed results and recipes outside the allowed calorie band. Saved recipes keep one per-serving nutrition record, so the recipe card, meal plan, and daily total read from the same values.
The percentage is one serving's stored nutrition divided by your daily target. If today's logged food is included, the ring shows today's total plus that serving.
The percentage is the sum of the day's resolved recipes divided by your daily target. If any recipe is missing complete nutrition data, the app labels the total as partial instead of silently treating the missing values as zero.
In a July 2026 engineering probe using a fixed vegan, pantry-free scenario and one reported daily target set, three generated plan runs landed at 97–99% of the calorie target, 97–104% of protein, 91–94% of carbohydrates, and 106–112% of fat. The reproducible probe and its result record are retained with the July macro-accuracy change.
97–99%
Calories
97–104%
Protein
91–94%
Carbs
106–112%
Fat
What this does not prove: this was a small internal engineering check, not an independent clinical validation or a guarantee that every generated meal will fall in these ranges. Different diets, pantry constraints, substitutions, and model output can produce different results.