Skip to main content

Glossary

The AI vocabulary, decoded.

Short, plain-English definitions for every term you'll bump into when shopping for or shipping AI agents.

Agent Framework

Software that helps developers build, run, and orchestrate AI agents.

Agentic AI

AI that takes actions in the world, not just generates text.

AGI (Artificial General Intelligence)

Hypothetical AI matching or exceeding humans across most intellectual tasks.

AI Agent

An LLM wired to tools, memory, and a goal — capable of multi-step action.

Autonomous Agent

An agent that runs without per-step human supervision.

Chain of Thought

Prompting technique that asks an LLM to reason step-by-step before answering.

Computer Use

An AI's ability to control a computer's screen, mouse, and keyboard.

Context Engineering

Designing what information goes into an LLM's context to get reliable behavior.

Context Window

How much text an LLM can consider at once when generating a response.

Embeddings

Numeric vectors that represent the meaning of text, images, or other data.

Evals

Tests that measure an AI system's quality on real tasks.

Few-shot / Zero-shot

Whether you give the LLM examples (few-shot) or just instructions (zero-shot).

Fine-tuning

Adapting a pre-trained LLM by continuing training on a smaller, task-specific dataset.

Foundation Model

A large general-purpose model trained on broad data, intended to be adapted to many tasks.

Function Calling

An older name for tool use — getting an LLM to invoke defined functions.

Guardrails

Rules and checks that prevent an AI system from saying or doing harmful things.

Hallucination

When an LLM produces output that's confident-sounding but factually wrong.

Inference

The act of running a trained AI model to generate output for a given input.

Inference Cost

How much it costs to run an LLM inference call.

LLM (Large Language Model)

A neural network trained on huge text corpora to predict and generate language.

MCP (Model Context Protocol)

Open protocol for connecting AI agents to external tools and data.

Multimodal

An AI model that can process and generate multiple types of input — text, images, audio, video.

Prompt Engineering

Designing inputs to an LLM to get reliable, high-quality outputs.

RAG (Retrieval-Augmented Generation)

Giving an LLM relevant context fetched from your data at query time.

Reasoning Model

An LLM trained or designed to spend extra compute on hard problems before answering.

RLHF (Reinforcement Learning from Human Feedback)

A training technique that aligns LLMs with human preferences using ranked outputs.

System Prompt

Top-of-conversation instructions that establish the model's role and rules.

Token

The basic unit an LLM reads and generates — roughly a fraction of a word.

Tool Use

An LLM's ability to call defined functions to fetch data or take actions.

Vector Database

A database optimized for similarity search over high-dimensional vectors.

Vibe Coding

Coding by describing what you want and letting AI generate it, without reviewing every line.