KI‑Assistent 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 real version of an AI effort that is already used in daily work and delivers a clear value.
- AI prototype An AI prototype is a fast, deliberately unfinished version of an AI idea that makes an assumption visible and testable.
- Artificial intelligence Artificial intelligence is the umbrella term for software that solves tasks which once required human judgement, such as understanding language, spotting patterns or generating suggestions.
- Deep learning Deep learning is a branch of machine learning that uses multi-layer neural networks to capture complex patterns in language, image or sound.
- Generative AI Generative AI is software that produces new content in response to an input, such as text, images, code or summaries.
- Large language models (LLM) A large language model (LLM) is an AI model trained on huge amounts of text that understands and produces language.
- Machine learning Machine learning is a subfield of AI in which a system learns from examples rather than having every rule prescribed.
- Multimodal AI Multimodal AI is a system that processes several kinds of input at once, such as text, image, sound or tables, and combines them into one result.
- Narrow vs. general AI Narrow AI solves clearly defined tasks, such as understanding language, drafting texts or sorting requests.
- Neural networks A neural network is a computing model of many connected units that turn inputs into outputs step by step.
- Vibe coding Vibe coding describes development where you describe to an AI assistant in everyday language what should be built and the AI produces the code.
Agency and consulting
- AI adoption AI adoption describes how far a team actually uses AI tools in daily work, not just trials them.
- AI agency An AI agency is a provider that plans, builds and runs AI solutions for companies rather than only giving recommendations.
- AI change management AI change management is the deliberate support of people during the introduction of AI.
- AI consulting AI consulting is support in identifying, assessing and preparing sensible AI initiatives.
- AI hackathon An AI hackathon is a short, intense format in which teams build on real cases within one or two days.
- AI literacy AI literacy is the ability to use AI sensibly, safely and critically.
- AI readiness AI readiness describes how well prepared a company is to use AI in daily work.
- AI roadmap An AI roadmap orders selected cases into a sequence with timings and dependencies.
- AI strategy An AI strategy sets where AI should create value in the company and in what order.
- AI transformation AI transformation describes the longer path on which a company anchors AI firmly in work and decisions.
- AI use case An AI use case is a clearly bounded application in which AI delivers concrete value.
- AI workshop An AI workshop is a facilitated format in which a team finds cases together, clarifies basics and sets a first step.
- Build vs. buy Build vs.
- Proof of concept A proof of concept is a small, targeted test that answers a single question: does the approach hold for this case?
Technology and models
- Agent orchestration Agent orchestration is the coordinated interplay of several AI agents and tools so a task runs through reliably.
- Agentic AI Agentic AI is an approach where AI breaks a goal into sub-steps on its own, uses tools and knowledge and adapts its course from intermediate results.
- AI agent An AI agent is a system that does not only answer a task but works on it in steps: it plans, calls tools, reads data and carries out actions until a goal is reached.
- Context window The context window is the amount of text a language model can consider at once for a request, measured in tokens.
- Embeddings Embeddings are rows of numbers that capture the meaning of a text, so similar content gets similar numbers.
- Fine-tuning Fine-tuning means training an already trained language model further with your own examples so it follows a wanted style, format or behaviour.
- Foundation model A foundation model is a large, broadly trained AI model that serves as the basis for many tasks, from understanding text through summarising to coding.
- Function calling Function calling is a language model's ability to invoke a clearly defined function with matching values instead of only producing text.
- Hallucination A hallucination is an answer from a language model that sounds fluent and convincing but is factually wrong or entirely made up.
- Inference Inference is the moment a trained model processes a concrete request and produces an answer.
- LLMOps LLMOps covers the practices that keep an AI assistant reliably operated, monitored and improved.
- Model Context Protocol (MCP) The Model Context Protocol (MCP) is an open standard that describes how an AI assistant talks to external tools and data sources.
- Multi-agent system A multi-agent system consists of several AI agents that solve a task together, each taking a bounded part.
- Open-source models Open-source models are language models whose weights are openly available, so you can download, run and adapt them yourself.
- Prompt engineering Prompt engineering is the work of phrasing instructions to a language model so it reliably delivers the wanted result.
- RAG RAG is a method where a language model searches an own knowledge source for relevant passages before every answer and builds them into the reply.
- Tokens Tokens are the small building blocks a language model splits text into, usually word parts or short character sequences.
- Vector database A vector database is a store that keeps texts as embeddings and finds the most similar in substance at high speed.
Assistants and voice
- AI answering machine An AI answering machine takes calls when nobody picks up and understands what they are about.
- AI assistant An AI assistant is a system that helps you and your team in daily work by answering questions, preparing tasks and drawing on approved knowledge.
- AI phone assistant An AI phone assistant takes incoming calls, understands the request and settles it in conversation or forwards it precisely.
- Chatbot A chatbot is a program that answers messages through a chat interface, usually within a clearly bounded topic.
- Company brain A company brain is an organisation's internal knowledge brain: a shared layer that draws on approved documents, handbooks, tickets and notes and turns them into citable answers.
- Conversational AI Conversational AI is the umbrella term for systems that hold a dialogue with people in natural language, by text or voice.
- Copilot A copilot is an assisting companion that helps directly inside a tool or work step: it suggests, drafts, summarises and waits for the human's confirmation.
- Inbound automation Inbound automation means taking, understanding and ordering incoming requests over phone, email, chat or form with an assistant.
- Outbound qualification Outbound qualification means reaching out to contacts actively and sorting whether and how they fit the offer.
- Speech recognition Speech recognition, in English speech-to-text or STT, turns spoken language into written text.
- Text-to-speech Text-to-speech, TTS for short, turns written text into spoken language.
- Voice agent A voice agent is an AI assistant that understands spoken language and answers, usually on the phone or over a voice channel.
- Voicebot A voicebot is a voice dialogue system that processes spoken input and answers by voice, often along predefined paths.
Automation
- AI automation AI automation combines automation with language models, so a flow does not only follow fixed rules but also understands, sorts and phrases unstructured content.
- API integration An API integration connects two systems through a dedicated interface, so they can exchange data and trigger actions directly.
- Exception handling Exception handling is the planned way of dealing with errors and exceptions in an automated flow.
- Hyperautomation Hyperautomation describes the approach of systematically automating as many of a company's flows as possible by making several tools work together: workflow automation, RPA, AI and interfaces.
- Intelligent document processing (IDP) Intelligent document processing (IDP) reads content from documents, sorts it and hands it structured to a system.
- Make Make is a cloud tool for workflow automation with a visual interface where flows are drawn as connected modules.
- n8n n8n is a tool for workflow automation that lets you assemble flows from individual building blocks.
- Process automation Process automation maps a whole business process with several steps, participants and rules automatically, rather than just a single task.
- RPA RPA stands for robotic process automation.
- Webhook A webhook is an automatic notification one system sends to another as soon as a certain event occurs.
- Workflow automation Workflow automation links individual work steps into a fixed flow that runs from a trigger to a result without manual nudging.
- Zapier Zapier is a cloud service that connects two or more apps through simple if-then rules.
Governance and law
- AI audit An AI audit is the systematic review of an AI system for rule adherence, quality and traceability.
- AI compliance AI compliance means the use of AI aligns with all applicable rules, from laws like the GDPR and EU AI Act to internal policies and sector requirements.
- AI governance AI governance is the framework of rules, roles and controls a company uses to steer how AI is allowed, checked and owned internally.
- AI literacy obligation (Art. 4) The AI literacy obligation under Article 4 of the EU AI Act requires providers and deployers to ensure a sufficient level of AI literacy among the people who operate or use their AI systems.
- AI risk classes AI risk classes are the categories the EU AI Act uses to sort AI systems by their potential for harm.
- Bias in AI Bias in AI means systematic distortions that lead a system to treat certain groups, views or outcomes unequally.
- Data protection and AI Data protection and AI covers how personal data is handled when AI systems are used.
- Data sovereignty Data sovereignty means a company keeps control over where its data sits, who accesses it and how it is used.
- EU AI Act The EU AI Act is the European Union regulation that governs the use of AI systems by risk.
- EU hosting EU hosting means data and applications run on servers within the European Union, under EU law and without a needless detour through third countries.
- Explainable AI Explainable AI describes AI whose outputs are traceable for humans.
- Guardrails Guardrails are rules and checks that limit what an AI assistant may say and do.
- Human in the loop Human in the loop describes a pattern where a human decides, confirms or corrects at a defined point in the AI process.
- Prompt injection Prompt injection is an attack in which manipulated text overrides an AI system's instructions.
Data and knowledge
- Data pipeline A data pipeline is the governed path along which data travels from its source to a usable target.
- Data quality Data quality describes how dependable the knowledge is that an AI assistant answers from: current, correct, complete, free of contradiction and with clear origin.
- Document extraction Document extraction pulls the intended content out of files such as PDFs, scans or forms.
- Knowledge base A knowledge base is an ordered collection of knowledge that an AI assistant reaches to answer questions.
- Knowledge graph A knowledge graph is a structure that represents things as nodes and their relationships as connections.
- Retrieval Retrieval is the targeted fetching of the right knowledge passages for a question before the assistant answers.
- Synthetic data Synthetic data is artificially generated data that mirrors real data without copying it one to one.
- Unstructured data Unstructured data is content without a fixed schema: texts, emails, PDFs, minutes, presentations or notes.
Process and operations
- AI implementation AI implementation is the path from an idea to an AI assistant that runs reliably in daily work.
- AI monitoring AI monitoring is the ongoing observation of an AI assistant in operation.
- AI operation AI operation is everything that keeps an AI assistant running after launch.
- AI pilot An AI pilot is a deliberately small, real test of an AI assistant before a decision on broad use.
- AI ROI AI ROI sets the value of an AI assistant against what it costs.
- Baseline A baseline records how a workflow runs today, before you change anything.
- Business process management Business process management is the discipline of deliberately designing, steering and improving a company's workflows.
- Process mining Process mining reconstructs from the traces in your systems how a workflow actually runs.
- Scaling AI Scaling AI means extending a proven AI assistant from a small use to many users, teams or cases.
- Total cost of ownership (AI) Total cost of ownership for AI is the sum of all costs of an AI assistant over its lifetime.
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.
Terms A–Z
A
- Agent orchestration Technology and models
- Agentic AI: AI that pursues goals in steps Technology and models
- AI adoption Agency and consulting
- AI agency Agency and consulting
- AI agent: an assistant that acts, not just answers Technology and models
- AI answering machine Assistants and voice
- AI assistant Assistants and voice
- AI audit Governance and law
- AI automation Automation
- AI change management Agency and consulting
- AI compliance Governance and law
- AI consulting Agency and consulting
- AI governance Governance and law
- AI hackathon Agency and consulting
- AI implementation Process and operations
- AI literacy Agency and consulting
- AI literacy obligation (Art. 4) Governance and law
- AI monitoring Process and operations
- AI MVP Fundamentals
- AI operation Process and operations
- AI phone assistant Assistants and voice
- AI pilot Process and operations
- AI prototype Fundamentals
- AI readiness Agency and consulting
- AI risk classes Governance and law
- AI roadmap Agency and consulting
- AI ROI Process and operations
- AI strategy Agency and consulting
- AI transformation Agency and consulting
- AI use case Agency and consulting
- AI workshop Agency and consulting
- API integration Automation
- Artificial intelligence Fundamentals
B
C
D
E
F
H
I
M
P
R
S
T
V
Strategy call
Book a strategy call.
Book your strategy call directly in the calendar: Book a strategy call. Alternatively: info@ki-assistent.co.
Pick a slot, the call is booked