Artificial intelligence glossary

An AI agent is an autonomous software system that perceives its environment, reasons, and takes actions to achieve a goal — often using an LLM as its brain and tools (APIs, search, code) to act.

Using AI to automatically perform repetitive or cognitive tasks on behalf of a human.

The ability of an AI model to analyze, understand and generate images or videos.

A structured sequence of steps where one or more AIs perform tasks inside a business process.

An interface that lets an application call a service's functions — including an AI model — via HTTP requests.

A field of computer science that builds systems able to perform tasks normally requiring human intelligence.

An AI system that plans and executes a series of actions to reach a goal with little human input.

A conversational program that talks with users in natural language.

The AI discipline that teaches machines to interpret the content of images and videos.

The maximum number of tokens a model can consider at once across input and output.

An AI assistant embedded in a tool to suggest, complete or execute actions alongside the user.

A structured collection of data used to train, validate or evaluate an AI model.

A subfield of machine learning using deep neural networks to learn complex representations.

A process that trains a small "student" model to mimic a large "teacher" model, keeping quality at lower cost.

An embedding is a numerical vector representation of text, image, or audio data that captures its semantic meaning, enabling similarity search and machine learning on unstructured content.

Fine-tuning is the process of further training a pre-trained model on a smaller, task-specific dataset to specialize it for a particular domain, style, or task.

A family of AI models that can create new content (text, image, audio, video, code) from a prompt.

An AI response that sounds credible but is factually wrong or fabricated.

An LLM is a large language model trained on massive text corpora, capable of understanding and generating natural language across many tasks: writing, summarization, translation, code, reasoning.

A branch of AI where models learn patterns from data instead of following explicit rules.

An open protocol that standardizes how an AI model connects to external tools and data sources.

The ability of an AI model to understand and generate multiple modalities (text, image, audio, video) inside a single system.

A brain-inspired model made of layers of artificial neurons that progressively transform data.

The branch of AI dedicated to understanding and generating human language.

A natural-language instruction sent to a generative AI model to get a response.

Prompt engineering is the discipline of designing, structuring, and refining the instructions given to an LLM to obtain reliable, accurate, and well-formatted answers.

A technique that reduces the numerical precision of a model's weights to make it lighter and faster.

RAG is a technique that combines an LLM with a search engine over your own documents: the model retrieves the relevant information first, then generates an answer grounded in that information.

A learning method where an agent learns to act by trial and error, maximizing a reward.

Technology that automatically transcribes spoken language into written text.

Learning from labeled examples (input → expected output).

Technology that synthesizes a realistic voice from written text.

A unit of text (often a word fragment) that a language model consumes and produces.

A neural network architecture based on the attention mechanism, foundation of modern LLMs.

Learning from unlabeled data, where the model discovers structures by itself.

A database specialized in storing and searching embeddings by semantic similarity.