20 AI Agents Vocabulary: Model, Token, RAG, MCP, Agent Loop and 15 More

Some of these words sound fuzzy but they are easy once someone says them plainly. Here are the 20 agentic AI terms you keep hearing in meetings, posts and docs, in about two minutes, with an example for each. If you have never heard them, it's your lucky day. Pause on the notes, the examples are there. What's inside 0:00 Why these words matter 0:10 What the model is: model, prompt, system prompt, context window, token 0:40 How it acts: tool, tool call, agent loop, workflow, harness 1:10 What it draws on: retrieval (RAG), memory, MCP, multi-agent 1:40 How you keep it safe: guardrail, eval, trace, prompt injection, hallucination, human in the loop This is video 2 in my Agentic Engineering series, short handwritten explainers for technical and business people who want to understand AI agents without the hype. Video 1 is "What's an agent?", next up is how much an agent should be allowed to decide on its own I build and review AI agents for small engineering teams. The contract triage agent I use as the example in this series is open on GitHub: https://github.com/nicanor-korir/cont... Say hi on LinkedIn:   / nicanor-korir   #AIagents #agenticengineering #learnAI #LLM #AIforbeginners