AI glossary.
34 terms, explained simply.
For everyone who is not an expert, and for those who want to explain it to their team. Every term has its own address for linking.
Basics
- Artificial intelligence
- Programs that solve tasks that would otherwise require human thinking: understanding language, recognizing images, finding patterns, preparing decisions.
- Language model (LLM)
- The “AI brain”: a program that understands and writes text. The engine of an AI system. Examples: Claude, GPT, Gemini, Llama. Learn more
- Prompt
- The instruction to a language model. In an AI system, prompts are built into the workflows, nobody has to type them.
- Token
- A piece of text, usually part of a word. Language models compute and charge in tokens.
- Hallucination
- A false statement that a language model phrases convincingly. Remedy: looking things up in your own knowledge with source references and checking before anything takes effect.
- Short-term memory (context)
- What a language model sees in the current conversation. Afterwards it is gone, which is why a business needs long-term memory.
- Long-term memory
- The permanent knowledge store of an AI system: customers, documents, agreements, decisions. Every AI role draws on it. Learn more
- Open model
- A language model whose weights are freely available, such as Llama or Qwen. It can run on your own hardware and be further trained. Learn more
How AI works
- AI agency
- A service provider that helps businesses use AI: from consulting through implementation to operation. Learn more
- AI architect
- Plans, builds and runs a business’s AI system as a whole: which capabilities, which workflows, which interfaces, which data location. Like an architect for a house. Learn more
- AI system
- Language model plus memory, senses, hands and rules. It works within your workflows instead of just answering questions. Learn more
- AI agent
- A program that uses a language model to plan and carry out steps on its own, for example looking things up, writing, entering data.
- AI role
- A specialized AI agent with a clear task, such as accounting, sales or legal, with its own rules.
- Workflow
- A fixed sequence of work steps that runs by itself, for example inquiry, quote, invoice. Learn more
- Process automation
- Recurring work is done by software. With AI, also work that requires reading, understanding and writing. Learn more
- Autonomy level
- How independently a workflow runs: from manual work through suggestion and approval to fully automatic with self-improvement.
- Lookup (RAG)
- Before every answer, the language model searches your documents and names the source. No training needed, always up to date. Learn more
- Fine-tuning
- A language model is further trained on your texts so that it masters your technical language, your tone and your cases. Learn more
Technology & devices
- Interface (API)
- The connection between programs. Through it, the AI system reads and writes in your accounting, your calendar or your devices.
- Local AI (on-premise)
- The AI system runs on your own hardware on your premises. The data does not leave the business. Learn more
- Cloud
- Computing and storage in a data center, accessible via the internet. With us, in the EU on request.
- Edge AI
- AI on a small computer right on site, for example at the camera. Fast and works even without internet. Learn more
- Camera AI (computer vision)
- AI that understands images and videos: people, vehicles, damage, fill levels. Learn more
- Speech recognition
- Speech is turned into text, for example in the phone assistant or for work reports by voice. Learn more
- AI chatbot
- An assistant on your website, WhatsApp or messenger that answers questions and books appointments. Learn more
- AI phone assistant
- Answers calls with your company name, understands the request, books appointments and writes a note. Learn more
Law & security
- GDPR
- The EU General Data Protection Regulation. It governs how personal data may be processed, including by AI.
- EU AI Act
- The EU law on artificial intelligence. It classifies AI by risk and sets obligations for providers and users. Learn more
- AI literacy (Art. 4)
- Since 2 February 2025, companies that use AI must ensure sufficient AI literacy among their employees. Learn more
- Transparency obligation (Art. 50)
- If an AI speaks directly with people, it must be recognisable that it is an AI. Learn more
- High-risk AI
- AI in sensitive areas such as recruitment or lending. Strict additional obligations apply here.
- Approval
- A person confirms before the system carries out an action. You decide per workflow where this is necessary.
- Data processing on behalf
- A service provider processes data on your behalf. This requires a contract under Art. 28 GDPR.
- Professional secrecy
- The duty of confidentiality of, for example, lawyers, tax advisors and doctors. One reason for AI on your own server. Learn more