How BrewMind builds your recipe.

AI reads the bean; fixed rules set the limits. Four steps happen while you wait, and your ratings shape the next cup.

Why one recipe does not fit every bean

Every bean extracts differently.

Origin, altitude, density, process and roast all change how water draws flavour out. A recipe written for one bean is only a guess for the next, and the same guess every morning gives the same middling cup.

So BrewMind asks a different question. Not which recipe should I use, but what does this bean need?

Reading the bean

It starts from your bag: origin, altitude, process, roast and tasting notes, taken from the label, a listing or what you type.

Working out extraction

From those traits AI estimates how the coffee will extract, and sets the ranges that grind, temperature and bloom should stay inside.

Building your pours

Inside those ranges AI writes the whole recipe: grind, temperature, bloom, and every pour with its amount and timing. Kettle mode and your xBloom follow the same recipe.

Checking every value

The recipe is scored for balance and clarity, and rebuilt once if it falls short. Then fixed rules check each number against physical limits before you see it.

One recipe, two ways to brew

A BrewMind recipe is a sequence you can run, not a note to read. It runs the same way by hand or on a machine.

Kettle screen: For brewing by hand, the recipe becomes a live guide: the time so far, the pour you are on, how much water goes in and when the next pour starts. Nothing to work out while you hold a kettle.
xBloom: For the machine, the recipe is sent as an xBloom profile: grind, temperature and every pour. Open it in the xBloom app or XBRW++ and brew, with nothing to set by hand.

The way you brew changes. The recipe does not.

Physics or AI

Not every change needs AI. Some follow from physics and can be calculated on the spot; others change the brew enough that the recipe has to be written again.

Dose and ratioCalculated: More coffee needs more water to wet the grounds, so the bloom scales with the dose and the main pours fill the rest of the water. The change appears at once, with no wait.
Hot and icedAI: Ice melts into the cup, so an iced recipe needs a finer grind and different pours. That is a new recipe, so AI writes it again. Its reading of the bean is reused, which makes it quicker than the first recipe.

That line is about getting it right, not speed: a change physics can answer never waits for AI, and a change that needs AI is never guessed with arithmetic.

What happens after you rate

After you brew, tell BrewMind how it tasted: sour, just right or bitter. It keeps your rating for this bean, and the next recipe moves a little in that direction.

Recipe
Brew
Taste
Rate
Adjust

Sour: Thin and sharp: the coffee did not extract enough. The next recipe grinds finer, pours a little hotter and, if needed, uses less water for the coffee.
Bitter: Heavy and dull: it extracted too much. The next recipe grinds coarser, pours a little cooler and, if needed, uses more water.
Just right: The recipe worked. It stays as it is, and the rating tells BrewMind to keep it.

Kayon Mountain tasted sour

Saved. Next time: 23 clicks (was 24).

A slightly finer grind, so it extracts a little more.

It is a nudge, not a promise. Your next rating tells BrewMind whether the change helped. Brews timed on the kettle screen also teach it about similar beans, because they record how the cup was actually made.

Why it matters

Consistency: The same care for every bean and every brew. No off days.
Adaptability: A printed recipe cannot react to your cup. Your ratings move each bean closer to what you like.
Precision: Every number comes from the bean, inside ranges worked out for it, and passes fixed limits. Nothing is copied from a template.
Explainability: Why this recipe shows the reason behind each setting, so you can understand it and trust it.
Getting better: Each brew you rate is evidence. When the same bean comes back, what was learned shapes its recipe before it is written.

Technical detail

Dose and ratio changes are calculated on the spot: bloom water scales with the coffee, the main pours fill the rest.

Switching between hot and iced rebuilds the pours with AI, because ice changes the grind and timing.

When a recipe scores below the bar, the critique becomes instructions and the recipe is written once more. The better of the two is kept.

Your bean. Your recipe.

Coffee is not a product to optimise. It is a material to understand.

Start with the bag you have.

Start brewing