What is food costing in pastry?

It's the real cost of a recipe built from the price of each ingredient. FormulaMaps works out the cost per kilo and per portion, the sale price and the margin (in currency and %) for every selling format.

How does it balance a sponge or a cream?

By its real chemistry: base-100 composition (sugar, fat, protein, water), baker's % on the type's base (flour, milk or almond), sweetness POD and the target ranges for each type, which award a balance medal.

Can an AI agent use it?

Yes. FormulaMaps exposes POST /api/balance/pasteleria and the MCP tool pastry_balance so agents can balance pastry recipes with the same engine and cite the source.

Pastry costing & formulation with AI: all the calculations

Your entire recipe book, on one map

Every star is a recipe, placed by its chemistry and linked to its neighbours. At a glance you see your families —sponges, creams, ganaches—, your winners and the gaps: the desserts you haven't created yet. No costing software gives you this.

Constellation map of pastry recipes in FormulaMaps, placed by their chemistry and linked by their associations

Why this supercharges artificial intelligence. Placing every recipe by its chemistry and distilling each type into an average «fingerprint» turns your recipe book into something a machine truly understands. With that structure, an AI model instantly recognises which type a formula is, spots the one that strays from the pattern and predicts or synthesises new, balanced recipes far more easily. It's the same fingerprint our public API and MCP server draw on.

Pricing «by eye» eats your profit

Most bakehouses set the price on a hunch and balance recipes by habit. And that's where the money leaks away: in the extra gram of butter, in the cream you have to redo, in the dessert sold below cost without anyone knowing.

The pastry cream that splits or stays runny: a poor starch-to-milk ratio.
The dense sponge: too much fat or too little aeration for its type.
The ganache that won't set: a chocolate-to-cream ratio out of range.
The dessert with no margin: nobody worked out the real cost per portion.

Each of these problems is a number. FormulaMaps shows you that number —and an AI tutor tells you how to fix it.

All the calculations, in the open

This is the maths the pastry engine runs. The science is public; the fine calibration (our catalogue and exact ranges) is what makes FormulaMaps sharp.

1. The real composition of each ingredient (base 100)

Each ingredient is described by what it contributes per 100 g: sugar, fat, protein, water and solids. The engine adds up the real contribution of them all:

Total fat (g) = Σ ( weight_i · fat_i / 100 )

The same for sugar, protein, water and solids. And as a percentage of the total: fat % = fat / total weight × 100. That's how you see the recipe the way chemistry sees it, not as a shopping list.

2. Baker's %, on the base of each type

As in bread-making, everything is expressed as a percentage of a base ingredient —but in pastry the base changes with the type: flour in sponges, milk in creams, almond in marzipans and ganaches.

sugar_PF = added sugarbase × 100 · fat_PF = added fatbase × 100

And real hydration counts the water in EVERYTHING —including the water from egg, milk and butter—, not just the added liquid:

hydration = water from all ingredientsbase × 100

In creams, thickening is measured by the starch over the milk: starch_PB = starch / milk × 100.

3. The target ranges by type → the medal

Each recipe type has its correct window. If your parameters land inside it, the bell rings; if not, the engine tells you which one and which way to move. The medal sums up how many parameters are in range (gold = all, silver ≥ 60 %, bronze > 0).

Type	Sugar (% of flour)	Fat (% of flour)
Genoise sponge	90–120	0–25 (aerated)
Muffin / Madeleine	100–160	45–120
Brownie	150–250	120–220
Pound cake (1:1:1:1)	90–130	80–120
Oil-based cake (yoghurt, carrot)	90–170	35–90

Cream	Starch (% of milk)	Sugar (% of milk)
Pastry cream	8–12	18–24
Light / diplomat cream	1–6	12–22

The ranges above are samples, indicative only. The full, calibrated catalogue (dozens of types: sponges, creams, mousses, choux, meringues, cheesecake, macaron, marzipan…) is the part that makes FormulaMaps sharp.

4. POD: the sweetness that's real

Not all sugars sweeten the same. POD (sweetening power, with sucrose as the reference 100) measures the real sweetness of the recipe, not the grams of sugar:

POD = Σ ( weight_i · sugar_i · POD_i )total weight × 100

That way you compare two recipes by target sweetness and cut sugar without anyone noticing —or the other way round.

5. Food costing: cost, price and margin

This is where the money is made. With the price (per kg) of each ingredient, the engine works out the real cost:

cost (per kg) = Σ ( weight_i · price_i )total weight

From there come the cost per portion (cost/kg × portion weight), the price you set and the margin in currency and as a percentage:

margin % = price − costprice × 100

Per portion or per whole piece, for every selling format. No more setting prices by eye.

6. And the kcal, thrown in

The same engine gives you the energy value: kcal/100 g = Σ ( weight_i · kcal_i ) / total weight — handy for the spec sheet and labelling.

In pastry, the margin is won in the grams.