Tell us your idea

Glossary

AI terms, explained in plain language

You hear them in every conversation about AI: agents, RAG, MCP, human-in-the-loop. Here each term gets a few sentences on what it is, how it works and when it matters for your business.

28 terms, alphabetical

Agentic agency

An agentic agency is an agency that builds and runs AI agents in production that carry out work inside a company on their own, connected to its systems, with a human in control.

Read the explanation

Agentic workflow

An agentic workflow is a process in which one or more AI agents plan and carry out the steps themselves, instead of a fixed script that prescribes every step.

Read the explanation

AI Act (EU AI Regulation)

The AI Act is the European law that sets rules for the use of AI, from light transparency duties for everyday applications up to strict requirements for high-risk applications.

Read the explanation

AI agent

An AI agent is software that handles a task end to end: reading, looking things up in your systems, drafting and acting, with a human checking the sensitive steps.

Read the explanation

AI literacy

AI literacy is the ability of staff to understand what the AI they use can do, what it cannot and where the risks are. Since 2 February 2025 the AI Act requires organisations to ensure it.

Read the explanation

AI readiness

AI readiness is the degree to which a company is ready to use AI meaningfully: are the processes clear, is the data reachable and are the people prepared.

Read the explanation

AI scan

An AI scan is a short, structured analysis of a company's processes that points out where AI and automation pay back fastest, and in which order to start.

Read the explanation

API

An API (application programming interface) is the official entrance through which software talks to other software: fetching, creating or changing data without a human clicking.

Read the explanation

CRM

A CRM (customer relationship management) is the system in which you keep customers, contacts, agreements and sales opportunities, such as HubSpot, Pipedrive or Teamleader.

Read the explanation

ERP

An ERP (enterprise resource planning) is the system in which a company keeps orders, stock, purchasing, planning and finance, such as Microsoft Dynamics, Exact or AFAS.

Read the explanation

Fine-tuning

Fine-tuning is additional training of an existing AI model on your own examples, so it masters a specific style or task better. For most companies it is unnecessary: a good prompt plus your own data (RAG) gets further.

Read the explanation

Hallucination

A hallucination is an answer from an AI model that sounds convincing but is wrong: an invented price, a non-existent product, a date based on nothing.

Read the explanation

Headless commerce

Headless commerce separates the front of your web store (what the customer sees) from the back (catalogue, cart, checkout), so you can design the front freely and the same back serves multiple channels.

Read the explanation

Human-in-the-loop

Human-in-the-loop means a human has a fixed place in the process: the AI prepares, the human approves or intervenes before anything becomes final.

Read the explanation

Knowledge base (with AI)

An AI knowledge base is a searchable collection of your own documents, manuals and agreements, in which staff or customers ask a question in plain language and get an answer with a source.

Read the explanation

LLM (large language model)

An LLM is the AI model that understands and writes text, such as the models behind ChatGPT, Claude and Gemini. It is the engine under an AI agent, not the agent itself.

Read the explanation

MCP (Model Context Protocol)

MCP is an open standard through which AI models and agents access tools and data sources in a uniform way, so an integration is built once and usable by any agent.

Read the explanation

Middleware

Middleware is the intermediate layer that sits between your systems and manages the traffic: translating formats, queueing, retrying and keeping track of what was exchanged.

Read the explanation

Prompt

A prompt is the instruction you give an AI model: the question, the context and the rules. With an agent the prompt is largely fixed in advance, so it behaves the same every time.

Read the explanation

RAG (retrieval-augmented generation)

RAG is the technique where an AI model first looks up the relevant passages in your own documents and answers with those, so the answer is based on your information and has a source.

Read the explanation

RPA (robotic process automation)

RPA is software that imitates human clicks and keystrokes on screens to repeat a fixed task. It works as long as the screen does not change and understands nothing of the content.

Read the explanation

Shadow AI

Shadow AI is AI use by employees outside the organisation's view: pasting customer data into a free chatbot, having contracts summarised in a tool nobody approved.

Read the explanation

Single source of truth

A single source of truth means every piece of data, such as a price, a stock level or a customer address, is managed in one place and arrives everywhere else automatically from there.

Read the explanation

System integration

System integration is connecting separate systems, such as web store, ERP, accounting and CRM, so data flows between them automatically and is the same everywhere.

Read the explanation

Token

A token is the piece of text (roughly part of a word) an AI model counts in. It determines the price of every request and how much text a model can process at once.

Read the explanation

Vector database

A vector database stores text as series of numbers that capture its meaning, so you can search for "what resembles this" instead of only exact words.

Read the explanation

Webhook

A webhook is a signal one system sends to another the moment something happens, for example "there is a new order", so the receiving system does not have to keep asking.

Read the explanation

Workflow automation

Workflow automation lets fixed steps between systems run automatically: when an order comes in, the invoice is created, the confirmation goes out and stock is updated, without anyone clicking.

Read the explanation

Rather see what it does in your business?

The AI Scan points out at least eight concrete opportunities in one afternoon, with a starting order. No theory, just your processes.

Book the AI Scan Read the blog

Chat on WhatsApp