AI Glossary for Builders

An AI program that processes your input and generates responses. Think of it as the engine powering your AI product.

AI models specifically designed to understand and generate text. Most can now also handle images and voice.

Your secret password to access AI services. Never share it or commit it to code!

The units AI models use to measure text - roughly 3/4 of a word. You're charged per token.

The instructions you give to an AI model. Good prompts = good outputs.

Controls how creative vs predictable AI responses are. 0 = deterministic, 2 = very creative.

How much text an AI can 'remember' in a single conversation. Measured in tokens.

When AI confidently states false information as fact. A critical issue to address.

Giving AI access to your specific data/documents so it can provide accurate, relevant answers.

Converting text into numbers that AI can understand and compare for similarity.

Training an AI model on your specific data to specialize its responses.

The hidden instructions that define your AI's personality, rules, and behavior.

Showing AI responses word-by-word as they're generated, like ChatGPT does.

Building AI products by iterating quickly with natural language instead of writing traditional code. You describe what you want, and AI generates the implementation.

Restrictions on how many AI requests you can make per minute/day.

A product built on top of existing AI APIs, adding your unique value proposition.

When users try to override your instructions to make the AI misbehave.

Letting AI trigger specific actions in your code, not just generate text.

AI that can work with text, images, voice, and other formats in one model.

How long it takes to get a response from AI. Every millisecond counts.

An open protocol that lets AI assistants connect to external data sources and tools through a standardized interface.

Switching from one AI provider to another (e.g., OpenAI to Anthropic).