The complete learning path inside the community — from the fundamentals to running your own AI workforce. No fluff, no filler. Here's exactly what you get.
The mental model. By the end you can read any AI hype post and tell what is real, what is buzzword, and what is worth your time.
Why most people are about to misuse AI for the rest of their career.
How LLMs actually think, without the buzzword soup.
Messy inputs in, useful signals out. The unsexy half of every AI workflow.
What an API actually is, and why every interesting integration starts here.
Why chatbots suddenly “get smart” when you wire them to your own docs.
The invisible ceiling every AI app hits, and how to design around it.
The moment AI stops being a search box and starts being a coworker.
11 episodes on how to think before you build. The frame that decides whether your automation actually saves you time, or just looks busy.
The pattern under every dead AI side project.
A simple layer model for deciding what AI should and shouldn’t touch.
Why your boring spreadsheet beats a smart agent every time you skip this step.
Drawing the workflow on paper before you let any AI near it.
Break it down until each step is so small AI can’t fumble it.
The decisions you should never hand to a model.
The five jobs models do better than humans — and the ten where they’re worse.
Cuts through the hype. Sets the bar for what counts as one.
The same six failure modes show up in every project. Spot them early.
A 20-minute decision framework so you stop building the wrong thing.
The endgame: running a real business with AI as the team.
n8n from zero. By the end you can wire your tools together with triggers, credentials, and core nodes — without writing code.
Why automations beat productivity tips, every time.
The wake-up call that fires every workflow.
OAuth without the rabbit hole. Connect your tools the right way once.
The shape of every API response — read it once, never panic again.
Talk to anything on the internet. Even tools n8n hasn’t officially integrated.
If, Switch, Filter, Merge — the seven nodes that handle 90% of real workflows.
A real two-part build, mapped and shipped end-to-end.
Self-host on a $5 VPS. Stop paying per execution.
Drop AI into the n8n workflows you already build. Extractors, classifiers, agents, RAG, and multi-agent systems — wired to your real tools.
The shift from “automate steps” to “automate judgment.”
Seven nodes that turn any workflow into a thinking workflow.
Four real systems. Inbox, content, data entry, error handling. All shipped.
Tools, prompts, sub-workflows. Build agents that do, not just answer.
Give your agents memory. Retrieval that actually finds the right doc.
A team of small agents beats one big one. The handoff patterns that work.
Where to put the human gate so the AI ships, but doesn’t embarrass you.
The standard that lets agents reach into any tool, no glue code.
Ship a real full-stack AI app without writing code from scratch. By the end, your app is live on a custom domain with auth, memory, and a vector-powered chatbot.
See the finished app on day one. Reverse-engineer how to get there.
Cursor, Github, Vercel, MCP. Live URL by lesson four.
The 3-layer mental model. Then build the chat that proves you understand it.
Stop shipping demos. Start shipping apps that remember who logged in last Tuesday.
Errors stop being scary once you know the loop.
Plug in your transcripts, docs, knowledge. The chatbot now knows your stuff.
When the chatbot answers, it shows the exact moment in the video.
Security, custom domain, the checklist you actually need before you ship.
The original primer. The eight building blocks of Claude Code, then a real ads-competitor pipeline you ship as your first skill.
Where Claude Code sits, and why the old IDE workflow is over.
The single feedback loop the whole tool is built around.
The interface in seven minutes. No yak-shaving.
CLAUDE.md — the one file that teaches the agent your project.
Memory that persists between sessions. The agent stops forgetting.
Skills — reusable instruction packs the agent picks up automatically.
MCP — connect Claude to any tool with one config line.
Permissions and settings so the agent can act without you babysitting.
Hooks — fire shell commands at lifecycle events. The duct tape that turns Claude into a daemon.
Eight blocks, one fluent system.
Bridge the agent to your n8n stack via MCP.
A real five-lesson build: scrape competitor ads, generate yours, ship a side-by-side site.
The deepest dive I’ve shipped — from your first CLAUDE.md to running a multi-employee AI office. Wire Claude into your real tools, build skills that ship work, go autonomous, then hire and coordinate a team of AI employees with the Glitch workforce.
How the course works — and a note for Windows users.
The shift from using tools to running teams.
The folder structure and CLAUDE.md that teach the agent your project.
Connect Claude to the stack you already run — Workspace, CRM, Stripe.
Turn Claude into a content engine: images, carousels, branded templates.
Editing, voice, research, and a working AI receptionist.
The one-job rule, a real end-to-end skill build, and the marketplace.
Loop, scheduled tasks, cloud routines — work that runs while you sleep.
The daily and weekly rhythm — and knowing it’s working.
What Glitch is, how it’s different, and getting it installed.
Meet the meta employee and hire your first AI employee by talking.
Migrate the skills you built into full AI employees.
Coordinate a multi-employee team from the Command Deck.
Extra builds — live dashboards and an integrated LLM wiki.
Full access to every course above, the community, and new drops every week.
Start building today
Professor Glitch