The terms behind AI agents, explained without technical jargon.
AI is moving fast, and so is its language. Here we collect the terms you meet when your organization puts AI assistants and AI agents to work – from language models and protocols such as MCP and KCP, to governance concepts like human-in-the-loop and AI Governance.
AI glossary – the terms explained without jargon | Mynder
AI assistant, AI agent, MCP, KCP, RAG and agentic AI – the glossary explained without technical jargon. Updated continuously by Mynder.
Technology that can analyse information, recognise patterns, produce content and solve a range of tasks.
AI that can create new content, such as text, images, audio, video and program code.
A large language model. This is the technology behind many modern AI assistants and agents.
A tool that helps you perform tasks, typically by answering questions, analysing information or producing content.
An AI-based system that can plan and perform actions to reach a goal, with access to specific tools and systems.
A term for AI systems that can work more autonomously, carry out multiple steps and use tools to solve tasks.
The instruction you give an AI model.
A technical way for different software systems to communicate with each other.
An open standard that makes it easier for AI systems to connect to external tools and data sources.
Think of MCP as a standardised connection between AI and the tools it uses.
A newer specification for describing and organising knowledge so AI systems can more easily find relevant information and understand which sources they should use.
KCP is still a newer approach and not as established as MCP.
A method where AI retrieves information from external sources before formulating an answer.
A protocol that makes it possible for different AI agents to communicate and collaborate.
The agent's training folder: defined, reusable recipes that teach the agent how a task is done, often bundling instructions, examples and tools in one place. That is knowledge.
Skills describe what the agent knows how to do – not what it is allowed to do.
The agent's access card and permissions: which systems it can enter, which data it can see and which actions it is actually allowed to perform. That is access and permission.
A capability means being allowed to and having access. Without the right capability, the agent's skills remain proposals and advice.
Functions an AI agent can use to act beyond the conversation, such as searching, reading files, sending email or updating a system.
The rules for which actions an agent may observe, propose or execute, and who must approve.
How much text and information an AI model can hold “in mind” at once during a task.
A log showing what an agent did, when, on what basis and who approved it.
When a task is performed automatically, without a human having to carry out each step.
AI agents can extend automation to tasks that require more interpretation and flexibility.
When AI produces information that sounds correct but is wrong or made up.
When a human must review or approve specific decisions or actions.
Systems, rules and processes that ensure AI is used responsibly, safely and verifiably.