AI Vocabulary Essential for Small and Medium-Sized Enterprises
Implementing artificial intelligence in a business requires an understanding of specific terminology, as well as the concepts, methods, and tools that frequently come up in AI projects. With this AI glossary designed for SMB executives, we help you clarify this terminology and make more informed decisions.
We have also included the acronyms and abbreviations most commonly used in the course of our work. This glossary will be expanded over time.
Direct access to the glossaries, listed alphabetically
AI Glossary from A to B
AI Act (European AI Regulation)
The European Union’s regulatory framework governing the design, marketing, and use of AI systems based on their level of risk, with stricter requirements for sensitive uses.
AI Agent (Autonomous Agent)
An AI system capable of performing a sequence of actions in a semi-autonomous manner (reasoning, planning, using tools) to achieve a defined goal, beyond a simple response.
API (Application Programming Interface)
A technical interface that enables two software applications to communicate with each other to exchange data and trigger actions (e.g., connecting a website, a CRM, and an AI tool).
Big Data
A very large, diverse, and often continuously generated dataset that requires specific tools to extract useful trends and insights.
AI Glossary: C–D
Chatbot
A conversational tool that automatically answers users' questions using a script or AI, in order to provide information, qualify requests, and guide users toward the appropriate action (contact, quote, purchase).
Connector
A module that connects an AI tool to an application or data source (CRM, Drive, Notion, website) to read or write information or automate actions.
Background
All the information provided to the model (instructions, history, relevant data) that enables it to generate a relevant and coherent response.
Data Privacy
All principles and measures designed to protect personal data (collection, storage, access, use), particularly when processed using AI tools.
Deep Learning
A subfield of machine learning based on deep neural networks, used to address complex problems such as language understanding, computer vision, and content generation.
AI Glossary: E through N
Embeddings
Digital representations (vectors) that capture the meaning of a text, a word, or a document to enable semantic search, comparison, and ranking by similarity.
Hallucination
A response generated by AI that seems plausible but is false, unverifiable, or fabricated, often due to a lack of reliable sources in the context.
Generative AI
A family of models capable of generating original content (text, images, audio, code) based on instructions, using training data.
LLM (Large Language Model)
An AI model trained on large amounts of text to understand and generate natural language (e.g., writing, summarization, conversational assistance).
Machine Learning
A branch of AI in which a model learns from data to make predictions, classifications, or recommendations without manually coded rules for each situation.
NLP (Natural Language Processing)
A field of AI that aims to analyze, understand, and generate human language (e.g., text classification, information extraction, summarization).
AI Glossary: O–R
Prompt
Text or instructions provided to a model (such as ChatGPT) to guide its response or content generation.
Prompt Engineering
A practice that involves designing clear, structured prompts to elicit responses that are more relevant, more reliable, and better aligned with a specific need.
System Prompt (System Instructions)
"High-level" guidelines that define the model's role, tone, rules, and priorities for an entire conversation or assistant.
RAG (Retrieval-Augmented Generation)
A method that combines a search of a knowledge base (internal documents) with AI-generated responses, in order to ensure that your content is more fact-based and better grounded.
Retrieval
A search step (often semantic) that involves identifying the most relevant documents or passages in a database before generating a response (often used in RAG).
GDPR and AI
Application of GDPR Rules to AI Projects: Legal Basis, Data Minimization, Transparency, Individual Rights, Security, and Management of Processors.
AI Glossary: S–Z
Temperature
A setting that controls the level of creativity/variability in responses: low = more stable and factual; high = more creative but riskier.
Tokens
Text units processed by models (word fragments/punctuation) that determine the “size” of a prompt and a response, and thus the cost and contextual limits.
Transcription (Speech-to-text)
Technology that automatically converts an audio file or speech into text (e.g., meeting minutes, voice memos).
TTS (Text-to-Speech)
Technology that converts text into synthetic speech, which is useful for voice assistants, audio content, and accessibility.
Vector database
A database designed to store and search embeddings in order to quickly find content that is “semantically similar” (semantic search, RAG).