KI-Automatisierung AI glossary
Overview
This editorial collection explains 100 AI terms the way they are used in real projects: with definition, use fields, industries and an honest decision guide. Every term links the relevant services and is published only after evidence, internal links, FAQ and bilingual parity are checked.
Fundamentals
- AI MVP An AI MVP is the smallest operable product version that tests one core value hypothesis with real users and data.
- AI prototype An AI prototype is a deliberately simple, working version of an AI idea that makes a central assumption testable.
- Artificial intelligence Artificial intelligence is software that solves tasks that would otherwise need human judgment, such as interpreting language, spotting patterns or deriving suggestions.
- Deep learning Deep learning is a subfield of machine learning that works with deep neural networks of many layers.
- Generative AI Generative AI describes models that produce new content instead of only sorting or scoring existing content.
- Large language models Large language models, or LLMs, are AI models trained on very large amounts of text that understand and produce language.
- Machine learning Machine learning is an approach in which a model learns patterns from data instead of someone programming every rule by hand.
- Multimodal AI Multimodal AI processes several input types in one model, such as text, image, audio or video together.
- Narrow vs. general AI Narrow AI solves a clearly bounded task, such as categorizing text or recognizing images, without general understanding.
- Neural networks A neural network is a computing model made of many simple, connected units that transform inputs step by step into an output.
- Vibe coding Vibe coding describes building software by telling an AI tool in natural language what should be created and then adjusting the result.
Agency and consulting
- AI adoption AI adoption describes how strongly an AI-supported flow is actually used in daily work.
- AI agency An AI agency plans and builds AI-supported workflows and products for companies that do not want to build a team for it.
- AI change management AI change management is the guided introduction of new, AI-supported flows in a team.
- AI consulting AI consulting means guiding a team to spot, assess and implement sensible AI uses.
- AI hackathon An AI hackathon is a short, focused format where a team builds a running AI prototype in one or two days.
- AI literacy AI literacy is the ability of people and teams to use AI sensibly and safely.
- AI readiness AI readiness describes how well a company is set up to use AI.
- AI roadmap An AI roadmap orders AI efforts into a sequence with clear interim goals.
- AI strategy An AI strategy defines what a company uses AI for, in what order and toward what goal.
- AI transformation AI transformation is the longer-term reshaping of a company toward flows that use AI as a matter of course.
- AI use case An AI use case is a concretely described application where AI adds clear value.
- AI workshop An AI workshop is a facilitated format where a team jointly explores fields for AI, assesses them and settles a first use case.
- Build vs. buy Build vs.
- Proof of concept A proof of concept (PoC) is a small, tightly bounded test that shows whether an AI idea works technically and in practice.
Technology and models
- Agent orchestration Agent orchestration is the coordinated control of several AI agents that solve a task together.
- Agentic AI Agentic AI describes an approach where AI systems plan goals on their own, break them into steps and carry them out via tools, often with several cooperating agents.
- AI agent An AI agent is a program that uses an AI model to pursue a goal on its own across several steps.
- Context window The context window is the maximum amount of text an AI model can consider at once in a single request, measured in tokens.
- Embeddings Embeddings turn texts, images or other content into number sequences that represent their meaning.
- Fine-tuning Fine-tuning further trains an existing AI model on your own examples so it solves a specific task more reliably.
- Foundation model A foundation model is a large, broadly trained AI model that serves as a base for many different tasks.
- Function calling Function calling is a language model's ability to output a structured function call with matching parameters instead of free text.
- Hallucination A hallucination is a language model output that sounds fluent and confident but is factually wrong or invented.
- Inference Inference is the process in which a fully trained AI model responds to a new input.
- LLMOps LLMOps covers the practices for operating, monitoring and improving applications built on language models.
- Model Context Protocol (MCP) The Model Context Protocol (MCP) is an open standard through which an AI model accesses external tools, data and systems.
- Multi-agent system A multi-agent system is a setup in which several AI agents, each with its own role, work together to solve a task.
- Open-source models Open-source models are AI models whose weights are openly available, so you can download, run and adapt them yourself.
- Prompt engineering Prompt engineering is the deliberate wording of instructions to an AI model so the output emerges reliably and in the desired format.
- RAG (Retrieval Augmented Generation) RAG (Retrieval Augmented Generation) connects a language model to a search in your own documents.
- Tokens Tokens are the small units into which an AI model splits text, usually word parts or frequent character sequences.
- Vector database A vector database stores content as number sequences called embeddings and finds the most similar entries for a query.
Assistants and voice
- AI answering machine An AI answering machine takes missed calls, records the message and turns it into text.
- AI assistant An AI assistant is a language-based tool that supports a person or team across several tasks, from research and summaries to simple actions in connected systems.
- AI phone assistant An AI phone assistant actively takes calls, understands the request and handles it in the conversation.
- Chatbot A chatbot is a text-based surface through which people ask questions and trigger tasks in dialogue.
- Company brain A company brain is a central knowledge layer that makes scattered company content from manuals, tickets, wikis and chats searchable.
- Conversational AI Conversational AI is the umbrella term for technologies that understand natural language and answer in dialogue.
- Copilot A copilot is an AI assistant that works directly inside a tool and makes suggestions while the human decides.
- Inbound automation Inbound automation processes incoming requests automatically, whether by call, email, chat or form.
- Outbound qualification Outbound qualification actively reaches out to contacts and checks whether real interest and fit exist.
- Speech recognition Speech recognition, also speech-to-text or STT, turns spoken language into written text.
- Text-to-speech Text-to-speech, short TTS, turns written text into spoken language.
- Voice agent A voice agent is a speech-driven AI assistant that in a conversation does not only answer but takes actions in connected systems.
- Voicebot A voicebot is a speech-based assistant that understands spoken requests and answers in dialogue.
Automation
- AI automation AI automation combines classic workflow automation with AI so that unstructured inputs like texts, emails or documents can be processed.
- API integration An API integration connects two systems through their programming interfaces so they exchange data directly.
- Exception handling Exception handling is the way an automated flow deals with exceptions, meaning everything that deviates from the normal case.
- Hyperautomation Hyperautomation describes the approach of systematically automating as many of a company's flows as possible and connecting several tools into one whole.
- Intelligent document processing (IDP) Intelligent document processing, IDP for short, reads content from documents and turns it into structured data.
- Make Make is a cloud-based automation platform where flows are built visually as a scenario of connected modules.
- n8n n8n is an automation platform where flows are built as a chain of nodes.
- Process automation Process automation hands an entire business process from start to finish to software, instead of replacing a single step.
- RPA RPA stands for robotic process automation.
- Webhook A webhook is a message a system automatically sends to a set address as soon as an event occurs.
- Workflow automation Workflow automation runs a fixed sequence of work steps without manual effort.
- Zapier Zapier is a cloud-based automation platform that works on an if-then pattern.
Governance and law
- AI audit An AI audit is the structured review of an AI system's function, risk, data handling and traceability.
- AI compliance AI compliance means an AI system verifiably meets the rules that apply to its use.
- AI governance AI governance is the set of rules a team uses to steer and take responsibility for its AI systems.
- AI literacy obligation The AI literacy obligation from Article 4 of the EU AI Act requires that people who operate or use AI systems have sufficient AI literacy.
- AI risk classes AI risk classes rank AI systems by their potential harm, from minimal through limited and high risk to prohibited applications.
- Bias in AI Bias in AI refers to systematic distortions in an AI system's results that unfairly disadvantage certain groups or cases.
- Data protection and AI Data protection and AI describes how personal data is handled in AI-assisted flows.
- Data sovereignty Data sovereignty refers to an organization's actual control over its own data, meaning who stores, processes, sees and may share it.
- EU AI Act The EU AI Act is the European Union regulation for artificial intelligence.
- EU hosting EU hosting means data and applications run on servers within the European Union, under EU law.
- Explainable AI Explainable AI describes AI systems whose decisions are traceable for humans.
- Guardrails Guardrails are fixed limits and rules that define what an AI step may do and what not.
- Human in the loop Human in the loop describes a flow in which a person reviews, approves or corrects an AI decision at a fixed point before it takes effect.
- Prompt injection Prompt injection is an attack in which manipulated inputs slip hidden instructions into an AI step.
Data and knowledge
- Data pipeline A data pipeline is an automated path along which data travels from a source to a destination, checked, reshaped and enriched on the way.
- Data quality Data quality describes how well data serves its purpose, measured by completeness, correctness, currency and consistency.
- Document extraction Document extraction is the automatic reading of relevant fields from documents such as invoices, contracts or forms.
- Knowledge base A knowledge base is an ordered collection of knowledge that a flow or model can access on purpose.
- Knowledge graph A knowledge graph is a structure that maps things and their relationships as a network of nodes and edges.
- Retrieval Retrieval is the targeted finding of matching content for a request before a model answers.
- Synthetic data Synthetic data is artificially generated datasets that reproduce the patterns of real data without representing real people or events.
- Unstructured data Unstructured data is content without a fixed schema, such as emails, PDFs, chat logs, images or free text.
Process and operations
- AI implementation AI implementation is the step in which a proven AI approach is built firmly into systems, data and workflows.
- AI monitoring AI monitoring is the ongoing supervision of an AI step in live operation.
- AI operation AI operation covers everything needed to keep an AI step running permanently after the start.
- AI pilot An AI pilot is a deliberately small, time-limited test of an AI step on a real use case.
- AI ROI AI ROI describes the ratio of value to cost of an AI automation.
- Baseline A baseline is the measured state of a process before a change.
- Business process management Business process management, or BPM, is the deliberate design, control and improvement of business processes.
- Process mining Process mining reconstructs the actual course of a process from the traces systems leave behind.
- Scaling AI Scaling AI means extending a working AI step from a single case to many cases, teams or sites.
- Total cost of ownership Total cost of ownership, or TCO, is the sum of all costs of an AI automation over its entire lifecycle.
Quality Standard
Every detail page has an extractable short definition, industries and use cases, a decision table, an example, related terms, evidence logic, one clear next step (book a call) and three questions with answers.
Book a strategy call
Book your strategy call directly in the embedded calendar Book a strategy call. Alternatively: info@ki-automatisierung.co.
Or send an email to info@ki-automatisierung.co