A map of AI as a field, in the order it actually builds on itself — not an alphabetical glossary. Each concept links to one well-matched course to start with.
Prefer a fixed, hour-by-hour course sequence instead? Use Learning path in the dashboard for that. This page is the "why" map; that one is the "what to watch, in what order."
Stage 1 · Foundations
The mental model everything else hangs off. Skip this and every later course spends half its runtime re-explaining things you'll wish you already knew.
What an LLM actually is
A large language model is a next-word predictor trained on huge amounts of text — it doesn't "know" facts the way a database does, it produces the statistically likely continuation of your prompt. That's why it can sound completely confident while being wrong (a hallucination), and why giving it better context changes its output more than arguing with it does.
Start with: Intro To Ai Ai Literacy For Everyone Beginner 2.9 hPrompting & prompt engineering
The instructions you give shape the output far more than the model itself. Prompting is the skill of being specific about role, format, constraints, and examples instead of vague requests — the single highest-leverage thing to get good at first, because it pays off in every tool downstream.
Start with: ChatGPT Prompt Engineering Essentials Beginner 4.7 hPicking and using an AI assistant day-to-day
ChatGPT, Claude, Gemini and Copilot all do the same core thing but differ in surfaces (chat, projects, artifacts, connectors) and strengths. Learning one deeply — Claude, in your case — beats shallow familiarity with all four.
Start with: Claude Mastery From Your First Prompt to Operating System Intermediate 4.1 hJudging AI output — literacy over blind trust
The most commonly skipped skill: knowing when an answer is probably right, when it needs a second check, and when the model is confidently making things up. This matters more as you hand it bigger tasks.
Start with: Ai Literacy For Employees Use Gen Ai Well At Work Beginner 1.3 hStage 2 · Extending what the model knows
A base model only knows its training data and whatever's in the current conversation. These are the techniques for feeding it your own information.
Context engineering
Distinguishes three ways to hand a model information: pasting it in the prompt, attaching it as a file, and connecting it to a live knowledge base. Picking the wrong one for the job is why some setups feel smart and others feel forgetful.
Start with: Beyond the Prompt Context Engineering for AI Intermediate 2.0 hRAG & vector databases
Retrieval-Augmented Generation is how you give a model access to a large private document set without retraining it: your documents get chunked, converted into embeddings (number vectors that capture meaning), stored in a vector database, and the most relevant chunks get pulled into the prompt at query time.
Start with: Databricks RAG Master Course Build Production AI Pipelines Beginner 1.7 hMCP — connecting AI to your tools
The Model Context Protocol is the standard that lets an AI assistant call out to real tools and data sources — your filesystem, a database, an API — instead of just talking. It's the plumbing behind almost every AI-agent integration you'll touch, so learning it once saves re-learning it per tool.
Start with: The Complete Model Context Protocol (MCP) Masterclass Intermediate 5.0 hStage 3 · Building with AI
Where you personally already live — Cursor and Claude are exactly this. Two genuinely different paths, kept separate on purpose since you only care about one of them.
AI coding assistants & vibe coding
Prompting an AI coding agent (Cursor, Claude Code, Windsurf, v0) to write and run real, working code for you — you're still shipping actual source code, just not typing every line by hand. This is your primary path.
Start with: Vibe-coding Basics Introduction to AI-Assisted Development Beginner 0.9 hNo-code AI app builders
Platforms like Lovable, Bubble, and Base44 that generate a working app from a description without exposing you to code at all — genuinely useful for non-developers, but a different skill from vibe coding since you never touch what gets produced.
Start with: Lovable 101 The Ultimate Beginners Guide Beginner 2.1 hStage 4 · Agents & automation
Once a model can call tools (via MCP) it stops being a chat window and starts being able to act — the next jump in capability, and where 'automate my own work' actually starts paying off.
What an AI agent is
An agent is a model given tools, memory, and the ability to decide its own next step in a loop, instead of just answering once. The key distinction from a plain chatbot: an agent can plan, act, observe the result, and adjust — that's also what makes it harder to predict and worth understanding before you rely on one.
Start with: AI Agents Build Your Mental Model Beginner 1.0 hMulti-agent systems & agent frameworks
Splitting a task across several specialized agents (a researcher, a coder, a reviewer) that hand work to each other, rather than one agent trying to do everything. Frameworks like LangGraph and the OpenAI/Claude Agent SDKs give you the coordination logic instead of hand-rolling it.
Start with: Claude Agents Build a Multi-Agent Research System in 7 Days Beginner 0.8 hAutomation platforms (n8n, Zapier, Make)
Visual, node-based tools for wiring triggers and actions together — often with an AI step embedded — without writing a full application. Complementary to agents rather than competing with them: many automations call an agent as one step in a larger workflow.
Start with: The Complete Beginner s Guide to N8N Beginner 0.7 hStage 5 · Governance & applying it at work
The last mile — using everything above responsibly, and translating it into value beyond your own workflow.
AI security & governance
Once AI agents can act on your behalf, new risks show up that don't exist with a plain chatbot: prompt injection, over-privileged tool access, and no audit trail for what an agent actually did. Worth understanding before you give an agent write access to anything that matters.
Start with: Responsible AI Ethics and Governance in the Workplace Beginner 1.0 hAI for leadership & management
A different lens on all of the above: not how to build with AI, but how to introduce it to a team, set expectations, and separate genuine value from hype when you're the one other people are asking for a recommendation.
Start with: AI For Business Leaders Turn Hype Into Responsible Advantage Beginner 1.0 hOnce you're comfortable across these, the Get certified track and the tag glossary go deeper on specific tools and vendor certifications.