Brewing + AI Guidebook

Glossary

Plain-language definitions of the AI and brewing terms used across this guidebook. Each term is tagged so you can see at a glance whether it comes from the world of AI Brewing or this Guide.

A

Attenuation Brewing

How much of the sugar in your wort the yeast converts to alcohol and CO₂. A common thing to predict, and a good test of whether an AI estimate matches what your strain really does.

AI (Artificial Intelligence) AI

Software that performs tasks usually needing human judgement, such as answering questions or spotting patterns. In this guide it usually means a chat-style assistant.

B

Bias (in data) AI

When your records lean one way and quietly mislead the tool, for example, if you only kept good notes when a batch went well, the data tells the model everything works.

C

Confidence AI

How sure an answer sounds or claims to be. Crucially, confidence is not the same as being correct, a model sounds equally certain whether it is right or guessing.

Context Guide

The brewing details you give the tool with a question, recipe, gravities, temperatures, kit, scale. More context turns a generic answer into a useful one.

D

Dataset AI

Your brewing records treated as data a tool can learn from or analyse. Its quality sets the ceiling on what any tool can do for you.

Diacetyl Brewing

A buttery or butterscotch off-flavor, often from rushed fermentation. Used throughout the guide as an example of something to catch early.

DMS Brewing

Dimethyl sulfide, a cooked-corn or cabbage off-flavor. One of the off-flavors a quality definition typically marks as an automatic fail.

F

FG (Final Gravity) Brewing

The gravity of the beer after fermentation finishes. Together with OG it gives you ABV, and it is one of the most common numbers an AI tool is asked to estimate.

G

Generative AI AI

AI that produces new text (or images) in response to a prompt, rather than just classifying things. The chat assistants this guide is about are generative AI.

H

Hallucination AI

When an AI confidently states something false, an invented figure, a made-up citation, a wrong rest temperature. A core reason to verify before acting.

House facts Guide

A short note of your brewery's real numbers and rules (mash efficiency, attenuation by strain, water profile) that you paste in to brief the tool so it answers in your terms.

L

Large Language Model (LLM) AI

The kind of AI behind chat assistants. The guide's mental model for it is a "very well-read apprentice" who has read widely but never tasted your beer.

M

Machine learning AI

Software that learns patterns from examples rather than being told every rule, like a brewer learning a yeast strain after a hundred batches, but faster.

Mental model Guide

Your working picture of what the tool is and is not. A clear one ("knowledgeable apprentice, not an authority") prevents both fear and over-trust.

Model AI

The trained AI tool itself. In this guide, the apprentice: fast and widely read, but with no taste, no memory of your brews, and no hands on your kit.

O

Off-flavor Brewing

An unwanted flavor or aroma such as diacetyl, DMS, acetaldehyde, or oxidation. Catching these early is a key promised use of AI in the guide.

OG (Original Gravity) Brewing

The gravity of the wort before fermentation, a measure of fermentable sugar. The starting point for predicting ABV and attenuation.

P

Pattern Guide

A reusable practice for using AI well in a brewery, like "give brewing context every time." See the Patterns page for the full set.

Prompt AI

What you type to ask or brief the tool. The guide's image is "how you brief the apprentice", the better the brief, the better the answer.

R

Reasoning AI

The working an AI shows behind an answer. Asking for it makes a claim far easier to check, and often makes the tool catch its own mistake.

Red flag Guide

A warning sign that an answer may be wrong, a precise number with no working, a confident claim about something recent, a citation you cannot find.

S

Sensory analysis Brewing

Evaluating beer by taste and aroma, often with a trained panel. The human judgement an AI tool supports but cannot replace.

SOP (Standard Operating Procedure) Brewing

A written, repeatable process. Often a better tool than AI for a routine job, and useful context to give the tool for everything else.

T

Training data AI

The examples a model learned from. It is fixed at a point in time, which is why a tool does not know this season's harvest or your latest batch.

U

Units Brewing

The scale a measurement is recorded in, Plato or specific gravity, Celsius or Fahrenheit. Mixing them in one field quietly breaks calculations and confuses tools.

V

Verification Guide

Checking an AI answer against the real brewhouse (your measurements, your experience, a small test batch) before acting on it.

No terms match your search.

Keep learning

Get new, practitioner-grounded pieces on AI and data for beer, wine, and whiskey by email, or read the full blog.

Subscribe by email Read the blog →

Bringing AI into your business? Talk to a brewing data scientist →