An 8-module AI bootcamp · narrated by my cloned voice
Stop watching AI tutorials.
Run the real system.
Most courses sell theory and a certificate. This hands you the working tools — the exact local AI, agent loop, and voice pipeline that keep a real autonomous operation alive. You don't watch me use them. You run them.
45s from Module 1 · built with the Module 6 pipeline
$ ollama run gemma3:12b # ...model loads locally. no key. no bill. $ gemma-code "write hello.py and run it" → wrote hello.py → ran: Hello, world — from an AI on your hardware. # that screenshot is your first deliverable.
You've watched the videos.
You still can't build anything.
You know the loop. Another "10 ChatGPT prompts" thread. A two-hour course that's a tour of a chat box. A certificate that proves you watched, not that you can do.
Meanwhile the meter runs. Every time you want to practice, you're paying per token to fumble — so you don't practice, so it never sticks.
The gap was never information. It's that nobody ever handed you the actual, running machine and said: here, drive it.
Why this one is different
Three things that are true here and nowhere else.
The system is the proof.
The course teaches itself.
Practice is always free.
Where you start. Where you finish.
Day 0 — today
- — AI is a chat box you rent by the message.
- — You can prompt, but you can't build.
- — "Agents" and "MCP" are words you nod at.
- — Every experiment costs money, so you don't.
By the capstone
- ✓ A local AI you own, with no subscription.
- ✓ A tool you built that the AI can call.
- ✓ A deployable product + an agent loop that runs while you sleep.
- ✓ Your own voice clone — and the judgment to use it right.
- ✓ A measurement loop you can point at anything.
Not a certificate. A running mini-operation you built, and the skills to extend it — proven, not theoretical.
The arc · zero → hero
8 modules. Each one ships something real.
~6 hours · run locally · $0 to practice
-
MODULE 0 · free
Your First Local AI
Install
ollama, pull a model, give it hands — an AI on your machine, offline, in 20 minutes.YOU BUILD → An AI that needs no subscription.
proof: the local fallback that kept a real op alive through a 2-day cloud outage.
-
MODULE 1
Prompting That Actually Works
Structure, not magic words. Role, context, constraints, format — and how to spot and kill AI-slop tells.
YOU BUILD → A reusable personal prompt library.
proof: the real humanizer + taste rules that keep the op's content from reading like a bot.
-
MODULE 2
AI in Your Terminal
Stop copy-pasting. Let AI read, write, and run your files — with guardrails so you review before it acts.
YOU BUILD → The ability to fix code by directing AI.
proof: this entire system is built and operated through an agentic terminal.
-
MODULE 3
Give AI Hands: Tools & MCP
AI that can do things — search, files, APIs — not just talk. Build your own tiny MCP server.
YOU BUILD → A ~20-line tool the AI can call.
proof: real MCP servers wired into the operation's toolbelt.
-
MODULE 4
Build a Real Product with AI
Scaffold a site, add lead capture with the real form-protection stack, learn the deploy basics.
YOU BUILD → A deployable 1-page site + working form.
proof: a real, byte-validated, multi-tenant product grown from one client site.
-
MODULE 5
Automate It: Your First Agent Loop
An AI that works while you sleep — scheduling, human-in-the-loop gates, and how to avoid runaway token-bleed.
YOU BUILD → A working automation with a human gate.
proof: the 24/7 swarm loop — and its real failure modes, taught honestly.
-
MODULE 6 · meta
Clone Your Own Voice
Record one clean clip, clone your voice locally, and learn the taste bar that keeps it from sounding like slop. Consent is covered explicitly.
YOU BUILD → Your own voice clone.
proof: this course is narrated by it — built with the exact pipeline you're using.
-
MODULE 7
The Self-Improving System
Instrument it, digest what happened, feed a critic, act on the highest-leverage fix. The loop that makes a system get better.
YOU BUILD → A feedback loop you can point at anything.
proof: the warehouse → digest → critic loop + the UTM→revenue attribution that runs this very course.
-
CAPSTONE ★
Your Mini Autonomous Operation
Combine all of it — local AI + a tool + a product + an agent loop + your voice + a measurement loop — into one small system that does something real for you.
YOU SHIP → A running operation, showcased in the Hive.
The part nobody else offers
Every exercise runs locally. For free. Forever.
Each module ships a folder you download and run on a free local model. No keys, no meter, no "you've used your free credits." Do the work as many times as you want — the cost of practicing is exactly zero.
Other courses paywall the doing. Here, the doing is the free part. You pay for the teaching and the working system — never for permission to try.
exercises/module-3/
- ├─ START_HERE.md
- ├─ prompts.md
- ├─ run.ps1 / run.sh
- └─ solution/
$ ./run.ps1
→ local model called. cost: $0.00
Credibility, the honest way
I'll teach you the wins. I'll also teach you the time it blew up.
Most courses hide their failures. The failures are the most valuable part. Two real ones you'll learn from directly — because I lived them:
Failure · taught in Module 5
The Reddit ban.
I pointed an automated loop at distribution. It posted like a bot, and the account got banned. The lesson is sharp and it's in the course: automation without a human and without taste doesn't scale you — it gets you removed. Be a contributor, not a dispatcher.
Failure · taught in Module 7
The distribution choke.
The content was good. Revenue was still zero — because the pipes were disconnected and I couldn't see it. That's the whole reason Module 7 exists: instrument everything, or you're flying blind. Measurement isn't optional; it's the difference between busy and working.
This is the proof the system is real: it has scars. Everything you'll learn points at a component that actually runs — the local fallback, the MCP toolbelt, the agent loop, the voice clone narrating this page, the measurement loop. Not a demo. The real thing.
Pricing
Start free. Pay once. Keep it forever.
Founding price — it goes up as the video build-alongs and live cohort come online. That's the only deadline, and it's a real one.
Module 0
Free
A real win, not a teaser.
- ✓ The full Module 0 lesson
- ✓ Its complete run-locally exercise
- ✓ A local AI on your machine in 20 min
- ✓ Access to Contrite's Hive (the community)
Full Course
$149 $199
Audio + slides. One payment, lifetime access.
- ✓ All 8 modules + the capstone
- ✓ Every exercise, prompt-pack & solution
- ✓ Lifetime access + all future updates
- ✓ Capstone rubric + showcase in the Hive
- ✓ 14-day "ship something or refund" guarantee
Course + Build-Alongs
$299 $399 opening soon
Watch the tools driven on screen. Honest: the build-alongs are being recorded now — reserve the founding price, pay nothing until they open.
- ✓ Everything in the Full Course
- ✓ Screen-recorded build-along per module (rolling out)
- ✓ Founding-cohort live Q&A (real, limited seats)
- ✓ Priority answers in the Hive
No subscription. No "make $X" promises — I sell the capability and the working system, nothing else. If you do Modules 0–2 and don't have a working local AI and a tool, email me inside 14 days for a full refund.
Honest answers.
Do I need to know how to code?
No. You need a Windows or Mac and the willingness to run a few commands. The whole point is that AI does the heavy lifting while you learn to direct it. Module 2 onward, you'll be reading and running code — but you're not writing it from scratch.
What does "free to practice" actually mean?
Every exercise runs on a free, open model on your own machine (ollama + gemma3:12b). No API key, no per-message billing. You can repeat any exercise as many times as you want and it costs nothing. You only pay once, for the course itself.
Is this really narrated by an AI clone of your voice?
Yes — built with the exact pipeline you learn in Module 6, with my own consent and my own reference recording. It's the proof the method works: the thing teaching you is the thing you'll build. Module 6 covers consent and ethics explicitly, because cloning a voice without permission is a hard line.
Why should I trust a course that brags about getting banned?
Because it's true, and because the failure is the lesson. A real system has scars. I'd rather you learn the agent loop and the way it can go wrong than buy a polished fiction where nothing ever breaks. Honesty is the differentiator, not a liability.
Will this make me money / get me a job?
I won't promise that, and you should distrust anyone who does. What I promise is the capability: by the end you'll have built a real, running mini-operation and have the skills to extend it. What you do with that is up to you.
What if it's not for me?
Start with the free Module 0 — get a local AI running before you spend a cent. If you buy the full course, do Modules 0–2, and don't end up with a working local AI and a tool, email me within 14 days for a full refund. The free module exists so you never have to take that on faith.
Get a local AI running
in the next 20 minutes.
Module 0 is free. No card. You'll have an AI editing code on your own machine, and a spot in the Hive where people post their first wins. Then decide if the rest is for you.
no email, no card · lesson audio + the run-locally exercise · ~4 MB .zip
Ready for all 8 modules? See pricing →