Glossary
AI glossary for business. Seven words you hear most often
Words that come up in every conversation about AI, explained the way we would explain them to a client over coffee. Each one with an example from real small business work.
Only the terms people actually search for get their own page. The rest are briefly at the bottom of this page.
Terms
- AI agent
AI that carries out several steps in your tools by itself, instead of only writing an answer.
- Prompt
The instruction you type into AI. Most of the answer’s quality depends on it.
- AI hallucination
An AI answer that sounds sure and is untrue. A made up number, rule or quote.
- Token
A piece of text, usually part of a word. AI counts chat length and limits in tokens.
- RAG
A way for AI to first find passages in your documents and then answer based on them.
- LLM
A large language model. The program behind ChatGPT, Claude and Gemini.
- AI Act
The EU regulation on artificial intelligence. It also covers businesses that only use AI.
Briefly about other words
- Model
- The program that generates answers. ChatGPT, Claude and Gemini are apps running models made by different companies.
- Context
- Everything the model sees in a chat, your messages, pasted files and its own replies. Context has a limit, which is why a long chat eventually goes wrong.
- Training
- Teaching a model on large collections of text. On personal accounts your chats may be used for it unless you switch that off in settings.
- Chat
- A conversation with AI in a window where you type and get text back. A chat suggests, but does nothing in your systems by itself.
- Connector
- A link between AI and another tool, such as a calendar or a drive. It lets AI read and write data there, so know what you give it access to.