The path
Six levels, non-technical → superpowered. A topic unlocks once its prerequisites are done.
L0Demystify
You are no longer intimidated by AI. You understand what a model actually is, what it can't do, and you can reliably get useful output from a chatbot.
- ✓12mWhat an AI model actually is
- ◐12mWhat AI can't (and shouldn't) doneeds: What an AI model actually is
- ◐15mTalking to AI: prompting that worksneeds: What an AI model actually is
- ·12mContext, memory, and why AI forgetsneeds: Talking to AI: prompting that works
- ·12mTrust but verify: hallucinations and checking outputneeds: What AI can't (and shouldn't) do, Talking to AI: prompting that works
L1Operate
You can produce real artifacts — documents, sites, automations — by directing AI tools, including vibe-coding tools, with confidence.
- ·12mThe everyday AI toolbeltneeds: Talking to AI: prompting that works
- ·12mFiles, projects, and giving AI contextneeds: The everyday AI toolbelt
- ·18mVibe-coding tools: describe it, get an appneeds: The everyday AI toolbelt
- ·15mDirecting AI to produce real artifactsneeds: The everyday AI toolbelt
- ·12mIterating: getting unstuck with AIneeds: Vibe-coding tools: describe it, get an app
L2Tinker
You are comfortable inside a terminal and a codebase that AI (or someone else) made — you can navigate, read, and not break things.
- ·12mWhat the terminal actually isneeds: Vibe-coding tools: describe it, get an app
- ·15mYour first real commandsneeds: What the terminal actually is
- ·12mFiles, folders, and what a repo isneeds: Your first real commands
- ·15mEnvironments, installs, and not breaking thingsneeds: Your first real commands
- ·18mReading code you didn't writeneeds: Files, folders, and what a repo is
L3Build
You can edit the code AI writes, understand the wider stack, and ship something live on the internet.
- ·18mThe wider stack: frontend, backend, deployneeds: Reading code you didn't write
- ·18mEditing the code AI writesneeds: Reading code you didn't write
- ·15mDatabases, conceptuallyneeds: The wider stack: frontend, backend, deploy
- ·15mDebugging with AI as a pairneeds: Editing the code AI writes
- ·20mShipping something liveneeds: The wider stack: frontend, backend, deploy, Editing the code AI writes
L4Orchestrate
You can build multi-step AI workflows and agents, give them tools, and judge the quality of what they produce.
- ·15mWhat an AI agent actually isneeds: Shipping something live
- ·18mTool use and MCP: giving AI handsneeds: What an AI agent actually is
- ·15mSubagents and delegationneeds: What an AI agent actually is
- ·18mChaining AI into workflowsneeds: Subagents and delegation, Tool use and MCP: giving AI hands
- ·15mEvaluating and trusting agent outputneeds: Chaining AI into workflows
L5Compound
AI is a durable force multiplier across your real work — you have your own toolkit, sound judgment, and a way to stay current.
- ·15mWiring AI into your real workneeds: Evaluating and trusting agent output
- ·15mBuilding your personal AI toolkitneeds: Wiring AI into your real work
- ·12mJudgment: when not to use AIneeds: Wiring AI into your real work
- ·12mStaying current without drowningneeds: Building your personal AI toolkit