Food identity
The food described by the user remains the starting point. A generic food and a specific packaged product are different kinds of reference.
TechnologymediaStrauß / Technology overview ↑
Food Intelligence
Food input is messy. Nutrition data should not be.
Food Intelligence is the technology behind KalorienKlar's ability to turn everyday food input into structured nutrition data.
People do not speak in database rows.
They write things like half a pizza, two rolls with sausage and a little butter, choose a packaged product or build a recipe from ordinary ingredients.
Food Intelligence is designed to keep the meaning of that input intact while connecting it with the best available nutrition reference.
Technology / How it works
From food language to nutrition data
A useful nutrition result starts with understanding what the person actually meant.
That includes the food itself, quantities, portions, preparation, additions and exclusions.
Food Intelligence structures that information before nutrition is calculated. Preserving what the user said comes before finding an easier match.
Quantities & Portions
Explicit quantities and units need to survive the move from ordinary language to structured data.
Components & Preparation
A meal can contain several ingredients, and preparation can change what a food description means.
Additions & Exclusions
What was added or left out is part of the input, rather than an optional detail.
Technology / Principles
Sources stay in charge
Where reliable food or product data exists, that source remains the basis for the result.
Food Intelligence helps connect real-world input with the appropriate reference and keeps track of where the nutrition data came from.
When an exact answer is not available, uncertainty should remain visible. The system should not create precision that the underlying information cannot support.
Source-backed nutrition
Nutrition values stay tied to their reference rather than becoming an unsupported model estimate.
Provenance
The origin of nutrition data remains relevant to how a result is understood.
Product identity
A specific product reference should remain distinct from a generic food reference.
Uncertainty
Missing information and explicit user quantities call for different treatment. Precision needs support from the available evidence.
Technology / How it works
Built for real food
Food is rarely as clean as a nutrition table.
A meal can contain several components. A product name can be incomplete. Quantities can be expressed in grams, pieces, glasses or ordinary language. Recipes combine ingredients with preparation and portions.
Food Intelligence handles the work between that messy input and the structured data a nutrition product needs.
Application / Real product
Built into KalorienKlar
KalorienKlar is the first product built on Food Intelligence.
The user sees a simple nutrition app. Underneath, the product has to understand food language, preserve quantities, resolve products and ingredients, work with recipes and keep nutrition tied to its source.
Food Intelligence is not a foundation model and it does not replace nutrition data.
It is the layer that helps KalorienKlar use that data reliably in real product situations.
KalorienKlar
Food Intelligence in a real nutrition product.
Technology / Continue exploring
Explore KINNO Intelligence