Prozessautomatisierung 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.
01
Fundamentals
- AI MVPAn AI MVP is the first lean but real version of an AI-supported flow that runs in production and delivers value.
- AI prototypeAn AI prototype is a deliberately small, quickly built test that checks whether an AI-supported flow works and delivers value.
- Artificial intelligenceArtificial intelligence describes software that solves tasks which usually need human judgement, such as understanding language, spotting patterns or proposing decisions.
- Deep learningDeep learning is a part of machine learning that learns from large data volumes with deep neural networks.
- Generative AIGenerative AI is a class of models that produces new content, such as text, summaries, reply drafts or code.
- Large language modelsLarge language models, or LLMs, are AI models trained on very large amounts of text that understand and generate language.
- Machine learningMachine learning is a subfield of AI in which a model learns patterns from example data instead of receiving every rule by hand.
- Multimodal AIMultimodal AI processes several input types at once, such as text, image, table and scanned document.
- Narrow vs. general AINarrow AI solves a clearly bounded task, such as reading receipts or classifying requests.
- Neural networksNeural networks are computing models loosely inspired by the brain.
- Vibe codingVibe coding describes a way of working where you describe software mostly in natural language and an AI model generates the code.
02
Agency and consulting
- AI adoptionAI adoption describes how strongly AI and automation are actually used in a company's daily work, not only introduced.
- AI agencyAn AI agency is a service provider that helps companies select, build and run AI solutions.
- AI change managementAI change management covers the measures with which a company embeds AI and automation into daily work instead of only introducing them technically.
- AI consultingAI consulting is the expert support for where and how AI creates value in a company.
- AI hackathonAn AI hackathon is a tightly time-boxed format in which mixed teams build on real automation cases, usually over one to two days.
- AI literacyAI literacy is the ability of people and teams to sensibly judge and use AI and automation and to recognise their limits.
- AI readinessAI readiness describes how well a company is prepared to use AI productively.
- AI roadmapAn AI roadmap is a time-ordered plan that puts automation and AI efforts into a sequence by value, effort and dependencies.
- AI strategyAn AI strategy defines which goals a company pursues with AI and in what order it gets there.
- AI transformationAI transformation is the longer-term reshaping of a company's flows, roles and decisions around AI and automation.
- AI use caseAn AI use case is a clearly bounded application in which AI or automation measurably improves a specific business flow.
- AI workshopAn AI workshop is a facilitated working format in which a team reviews, assesses and sequences automation potential together.
- Build vs. buyBuild vs.
- Proof of conceptA proof of concept, PoC for short, checks on a clearly bounded case whether a planned automation is feasible technically and functionally.
03
Technology and models
- Agent orchestrationAgent orchestration means coordinating several AI agents in one flow so that each takes a subtask while an overarching flow controls the order, handovers and limits.
- Agentic AIAgentic AI describes systems where AI plans, executes and coordinates tasks on its own across several steps or several cooperating agents to reach a goal.
- AI agentAn AI agent is a program that uses a language model to handle a task on its own across several steps.
- Context windowThe context window is the amount of text a language model can process at once, measured in tokens.
- EmbeddingsEmbeddings are strings of numbers that represent the meaning of a text, image or document in a form a computer can compare.
- Fine-tuningFine-tuning means training an already trained model further with your own examples so it handles a specific style, format or task more reliably.
- Foundation modelA foundation model is a large AI model pre-trained on broad amounts of data that serves as the basis for many applications.
- Function callingFunction calling is a language model's ability to call a predefined function with matching values instead of only producing text.
- HallucinationA hallucination is an output from a language model that sounds plausible but is factually wrong or invented.
- InferenceInference is the process where a trained AI model responds to a concrete input.
- LLMOpsLLMOps stands for operating applications with large language models in everyday use.
- 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 systemA multi-agent system is a setup where several specialised AI agents work together instead of one agent doing everything.
- Open-source modelsOpen-source models are AI models whose weights are openly available, so you can run and adapt them yourself.
- Prompt engineeringPrompt engineering is the craft of giving a language model instructions clear enough that it reliably delivers the wanted result.
- RAG (Retrieval Augmented Generation)RAG (Retrieval Augmented Generation) is a method where a language model searches a private knowledge source for relevant documents before answering and lets them flow into the response.
- TokensTokens are the small units into which a language model splits text, mostly word parts or short character sequences.
- Vector databaseA vector database is a store that keeps content as number strings (embeddings) and finds the ones most similar by meaning at high speed.
04
Assistants and voice
- AI answering machineAn AI answering machine takes calls when no one is available, captures the request and forwards it in a structured way.
- AI assistantAn AI assistant is software that responds to natural language and helps an employee complete tasks across several systems.
- AI phone assistantAn AI phone assistant is a speech-based service that takes calls on a business phone line, understands requests and handles them.
- ChatbotA chatbot is a text-based dialogue service that answers user messages in a channel such as a website, app or messenger.
- Company brainA company brain is a central knowledge layer that makes scattered company knowledge from documents, systems and data sources searchable and usable for AI.
- Conversational AIConversational AI is the technology category behind systems that understand natural language and answer it in a dialogue.
- CopilotA copilot is an AI-backed companion that runs right inside an existing work tool and helps the user with the current task.
- Inbound automationInbound automation captures incoming requests across channels such as phone, email, form or chat, understands the request and forwards or answers it automatically.
- Outbound qualificationOutbound qualification actively approaches selected contacts, checks their suitability against defined criteria and prepares fitting cases for the next step.
- Speech recognitionSpeech recognition, also speech-to-text or STT, turns spoken language into written text.
- Text-to-speechText-to-speech, or TTS, turns written text into spoken language.
- Voice agentA voice agent is a speech-driven AI service that runs a spoken conversation from start to finish while triggering actions in connected systems.
- VoicebotA voicebot is a speech-based dialogue service that handles spoken standard requests along defined paths.
05
Automation
- AI automationAI automation puts machine learning or language models at exactly the points in a flow where fixed rules no longer suffice.
- API integrationAn API integration connects two systems through their official interfaces so they exchange data and trigger actions in a structured way.
- Exception handlingException handling is the planned treatment of cases that deviate from the normal flow.
- HyperautomationHyperautomation describes the approach of systematically automating as many of a company's flows as possible by combining several techniques: RPA, AI, integration and process analysis.
- Intelligent document processing (IDP)Intelligent document processing, IDP for short, turns documents of all kinds into structured, usable data.
- MakeMake is a hosted platform for visual automation where you build flows as scenarios from connected modules.
- n8nn8n is a workflow automation platform where you assemble flows from prebuilt building blocks (nodes) without coding everything yourself.
- Process automationProcess automation means having software run recurring business flows fully or partly instead of doing them by hand.
- RPARPA, short for Robotic Process Automation, means a software robot runs fixed flows directly on the surface of an application.
- WebhookA webhook is a message a system sends automatically to a stored address as soon as a specific event occurs.
- Workflow automationWorkflow automation means letting a fixed sequence of tasks run on its own through triggers, conditions and actions.
- ZapierZapier is a hosted service that connects single apps through triggers and actions.
06
Governance and law
- AI auditAn AI audit is the systematic check of an AI flow for function, data, risks and evidence.
- AI complianceAI compliance is the adherence to the rules that apply to AI use, and its evidence.
- AI governanceAI governance is the set of rules, roles and controls with which a company steers its use of AI.
- AI literacy obligation (Art. 4)The AI literacy obligation under Art.
- AI risk classesAI risk classes are the tiers with which the EU AI Act divides AI systems by their risk.
- Bias in AIBias in AI is a systematic distortion in a model's results that treats certain groups or cases unfairly.
- Data protection and AIData protection and AI describes how the protection of personal data is upheld when AI is used.
- Data sovereigntyData sovereignty means a company keeps control over its own data, that is over where it sits, who accesses it and how it is used.
- EU AI ActThe EU AI Act is the European Union regulation that classifies AI systems by risk and attaches graded duties to that.
- EU hostingEU hosting means the data and services of a flow run on servers within the European Union.
- Explainable AIExplainable AI is the approach of making AI decisions traceable enough that people understand why a result came about.
- GuardrailsGuardrails are technical and organisational limits that define what an AI step in an automated flow may and may not do.
- Human in the loopHuman in the loop means a person checks, approves or corrects at a defined point of an automated flow.
- Prompt injectionPrompt injection is an attack where instructions hidden in input data make an AI model ignore its actual task.
07
Data and knowledge
- Data pipelineA data pipeline is a chain of automated steps that pulls data from a source, reshapes it and writes it into a target.
- Data qualityData quality describes how fit data is for its purpose: whether it is complete, correct, current, consistent and free of contradiction.
- Document extractionDocument extraction means automatically reading individual fields and content from documents and passing them on in a structured form.
- Knowledge baseA knowledge base is a maintained collection of checked knowledge that a flow or an AI step accesses.
- Knowledge graphA knowledge graph is a data model that represents concepts and objects as nodes and their relationships as edges.
- RetrievalRetrieval is the targeted fetching of matching information from a knowledge source before an AI model answers.
- Synthetic dataSynthetic data is artificially generated data that mirrors real data without containing real people or events.
- Unstructured dataUnstructured data is information without a fixed field schema, such as emails, PDFs, scanned receipts, contracts or free text.
08
Process and operations
- AI implementationAI implementation is the work of turning a tested idea into a running, productive flow.
- AI monitoringAI monitoring is the ongoing observation of a live AI step for quality, behaviour and cost.
- AI operationAI operation is the ongoing responsibility for a productive, automated flow after it has gone live.
- AI pilotAn AI pilot is a deliberately small, time-boxed test that tries an automation on a real flow before it is rolled out broadly.
- AI ROIAI ROI is the ratio between the provable value of an AI automation and the cost it causes across rollout and operation.
- BaselineA baseline is the measured starting line of a process before a change.
- Business process managementBusiness process management (BPM) is the end-to-end discipline of capturing, designing, running, measuring and continuously improving business processes.
- Process miningProcess mining reconstructs the actual flow of a process from the log data of the systems involved, instead of relying on drawn target diagrams.
- Scaling AIScaling AI means moving a proven AI automation step from the pilot to more volume, more flows or more sites.
- Total cost of ownershipTotal cost of ownership (AI) is the full cost of an AI automation across its whole lifecycle, not just the rollout.
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
Straight into the calendar
Book your strategy call directly via the calendar: Book a strategy call. Alternatively: info@prozessautomatisierung.co.
Or send an email to info@prozessautomatisierung.co