25 terms

AI glossary, in plain English

What the words mean, without the jargon. Each entry gives a one-sentence definition first.

A

agentic workflow

An agentic workflow is a repeatable process in which an AI carries out several connected steps, using tools and making small decisions, with human checkpoints at the important points.

AI agent

An AI agent is an AI system that pursues a goal by taking multiple steps on its own, such as reading files, using tools and checking its work, instead of only answering one question.

AI automation

AI automation is the use of AI to carry out work steps automatically, especially steps that involve messy language such as emails, documents and notes that fixed rules cannot handle.

API

An API (application programming interface) is a set of rules that lets one piece of software send requests to another, such as a script asking an AI model for an answer.

C

A chatbot is a program you talk to in a chat window, which today usually means an AI assistant that answers in natural language rather than a fixed menu of scripted replies. ChatGPT, Claude, Gemini and Copilot are examples. Older chatbots followed decision trees; modern ones use large language models to understand free-form questions.

Claude Code

Claude Code is Anthropic's agentic AI tool that reads files, makes changes and runs commands from a plain-English request, available in a terminal, IDE extensions, a desktop app and the web.

Codex

Codex is OpenAI's agent platform in ChatGPT that takes a goal, gathers context, takes action and produces a result, rather than only replying in a chat.

context window

A context window is the amount of text an AI model can hold in view at once, covering your instructions, the files you share, the conversation so far and its own replies.

Copilot

Microsoft Copilot is Microsoft's AI assistant, built into Microsoft 365 apps such as Word, Excel, PowerPoint, Outlook and Teams and available as a chat.

custom GPT

A custom GPT is a version of ChatGPT set up for one job, with saved instructions and, optionally, reference files, so it behaves the same way each time you use it. As of November 2026, OpenAI is transitioning custom GPTs to plugins, which package reusable skills, reference files and connected apps; retirement timing varies by workspace.

E

embeddings

Embeddings are lists of numbers that represent the meaning of a piece of text, image or other content, so software can find items that are similar in meaning.

F

fine-tuning

Fine-tuning is further training an existing AI model on your own examples so it adopts a particular style, format or specialty. To fine-tune a model means teaching it your pattern with hundreds of sample inputs and ideal outputs. Most teams do not need it: better prompts and your own documents come first.

G

generative AI

Generative AI is software that creates new content, such as text, images, audio, video or code, from a plain-language instruction instead of following a fixed script.

H

hallucination

An AI hallucination is a response that sounds confident and fluent but contains information that is false, invented or not supported by any source. It happens because language models generate likely-sounding text rather than look up verified facts. Fabricated statistics, fake citations and wrong dates are typical, and checking important claims remains your job.

L

LLM (large language model)

A large language model (LLM) is an AI system trained on huge amounts of text so that it can read, write, summarize and answer questions in ordinary language.

M

MCP (Model Context Protocol)

MCP, the Model Context Protocol, is an open standard that lets AI applications connect to external data sources, tools and workflows in a common way.

multimodal AI

Multimodal AI is an AI system that can work with more than one kind of input or output, such as text, images, audio, video and documents, in the same conversation.

O

OpenClaw

OpenClaw is an open-source, self-hosted gateway that connects messaging apps such as Slack, Telegram and WhatsApp to AI agents so you can message an assistant from apps you already use.

P

Perplexity

Perplexity is an AI-powered answer engine that searches the web and writes a response with citations to the sources it used.

prompt

A prompt is the instruction, question or material you give an AI tool so it knows what to do.

prompt engineering

Prompt engineering is the practice of writing, testing and improving the instructions you give an AI tool so that it returns reliably useful results.

R

RAG (retrieval-augmented generation)

RAG, or retrieval-augmented generation, is a technique where an AI first searches a set of trusted documents and then writes its answer using what it found.

reasoning model

A reasoning model is an AI model designed to work through a problem in steps, spending extra time thinking before it gives its answer.

S

system prompt

A system prompt is a standing set of instructions, given before the conversation starts, that tells an AI tool how to behave, what role to play and what rules to follow.

T

token

A token is the small chunk of text, often a word or part of a word, that an AI model reads and writes, and the unit its limits and prices are counted in. In English, a token is very roughly three-quarters of a word. Your prompt, any attached files and the reply all use tokens.

A