AI Glossary & FAQ
The most important terms from the AI world, explained clearly. Every technical term on this website leads here.
Glossary
Artificial Intelligence (AI)
Software that handles tasks which used to require human thinking: writing texts, generating images, recognising patterns, preparing decisions.
Machine Learning
A branch of AI: a model learns from examples instead of hard-coded rules. The better the examples, the better the result.
Large Language Model (LLM)
A model trained on vast amounts of text that understands and produces language. The foundation of ChatGPT, Claude and comparable systems.
Training
The learning process of a model. In a business context it usually means: familiarising the AI with your profile, wording and design so it can work in your company's name.
Prompt
The input given to an AI: a question, a task or context. Good prompts are precisely worded work orders.
Token
The smallest processing unit of a language model, roughly a syllable to a short word. Tokens are what models compute with, and often what you pay for.
Context Window
How much text an AI can keep in view at once. The larger it is, the more company knowledge fits into one session.
Retrieval-Augmented Generation (RAG)
The AI looks things up in your knowledge base before answering and replies using your documents instead of memory alone. This lowers error rates considerably.
Hallucination
A convincing-sounding but wrong AI answer. The remedy: a good knowledge base, retrieval (RAG) and clear guidelines.
AI Agent
An AI that carries out multi-step tasks on its own: researching, creating files, operating tools, delivering results.
Open-Source Model (Open Weights)
A language model whose weights are freely available. It can run on your own hardware, the foundation of sovereign AI.
On-Premises
Operation on your own servers instead of someone else's cloud. Maximum control over data and availability.
Sovereign AI
AI systems that a company or state runs under its own control: own infrastructure, own data, own rules, free choice of provider.
Vendor Lock-in
Dependence on a single provider because data and processes are trapped in its system. Open formats dissolve the dependency.
Knowledge Base
A company's structured, searchable body of knowledge: documents, rules, products, experience; in open formats, usable with any AI.
Design Tokens
Design decisions as machine-readable values: colours, typefaces, spacing. People and machines design with the same rules.
llms.txt
A text file on a website that explains the brand to AI systems: offer, tone, rules. The counterpart to robots.txt, for language models.
Model Context Protocol (MCP)
An open standard through which AI systems access tools and data sources, for example a brand rule set.
EU AI Act
The European AI law. Among other things it requires demonstrable AI literacy from staff since February 2025 (Article 4) and regulates AI by risk class.
General Data Protection Regulation (GDPR)
European data protection law. For AI, it decides where company data may flow and on what legal basis.
Frequently asked questions (FAQ)
Do I need to know how to code to use AI in my company?
You receive tools we have set up and a trained workforce. The AI is operated in plain language; programming skills are dispensable.
What does it mean when the AI “learns” our company?
We hand the AI your profile, your corporate wording, your design rules and product definitions in structured form. It then works in your tone and by your rules, see the page “Your AI learns your company”.
Does our data stay with us?
Yes, that is the core of our offer: operation on servers in Germany and the EU or on-premises, a knowledge base in open formats, provider switch possible at any time.
What does entry cost?
Sovereignty Check from €1,900, Lock-in Check from €1,490, one-day workshop including AI policy from €2,900 flat, complete machine-readable corporate design package from €4,900. The initial consultation is free.
Are we legally required to train our staff?
The EU AI Act has required demonstrable AI literacy since February 2025 in Article 4 for staff using AI. Our one-day workshop delivers training and policy in one.
Which AI is right for us?
That is what the Sovereignty Check clarifies: we assess your situation and goals and recommend an operating model, from an open-source model in an EU cloud to on-premises operation.
Does the AI replace our employees?
It accelerates them. Tasks that used to cost days, a trained corporate communication delivers in minutes; your employees gain time for the work that needs judgement.
Does this also work with ChatGPT, Claude and others?
Yes. Your knowledge base and your machine-readable brand rule set are built vendor-neutrally and usable with any common AI, today and at your next switch.