Stop chatting with AI. Start managing it.
Leave with a team of AI agents that use your real tools, remember what matters, and work while you sleep.
Live build sessions, three a day. 15 spots.
“You don't have a knowledge problem. You have a finishing problem.”
You've watched the agent demos. You've read the threads. What you've never had is a room with a deadline, a facilitator on your build every day, and a cohort that won't let you quit. In 10 days, that room turns “I chat with AI” into “my agents did that while I slept.”
A workforce. Running, not planned.
Every outcome is a hard deliverable with a day attached, verified by your pod and demoed in public.
3+ coordinated agents with defined roles. One researches, one writes, one reviews. An org, not a bot. You're the manager.
Connected to your email, calendar, and GitHub via MCP. Gated by the Rule of Two, with human approval on money, send, and delete.
Your agents remember you across sessions and run on a schedule, with traces and a cost number you can say out loud.
Day 10: your capstone, demoed live with receipts, plus a 30/60/90 plan so it keeps compounding after.
The 5-layer model. The map that never changes.
Tools change monthly. This model is what survives. Every day of the sprint climbs this stack.
You start at “I chat with AI” and climb one layer at a time, until you're running an org.
Two working weeks. One AI workforce.
No calendar gymnastics: Day 1 through Day 10, each one ending with something shipped.
The 5-layer model, repo literacy, and something of yours live on the internet today.
Turn a task you hate into a job an agent does. Your first reusable skill.
Claude Code and Codex. Tasks done while you sleep. What agents actually cost.
OpenClaw, SOUL.md, Hermes. And when not to use an agent at all.
MCP. Connect two or more real tools. Rule of Two verified before any private data.
An agent that remembers you. An Obsidian brain. Browser workflows.
The org-chart idea. A 3-agent workflow running end to end.
Roles, budgets, reporting. Specialist subagents. Red-team your own org.
Always-on hosting, guardrails, traces, and a cost number you know.
Your working workforce, demoed live and in public, with receipts.
🚢 Every single day ends with something shipped. No passive lectures, ever.
Three hours live. The rest is yours to build.
The why, the layer, the capability. Taught by building.
We ship today's piece together.
Homework pack, builder extensions, pod help, Agent Mentor.
Demo progress and a manage-your-worker reflection.
🚢 Evenings: post proof of today's ship in your pod channel. Every day.
Guardrails first. Power second.
The threat model, before any agent touches real data.
No tool touches private data until you can state and apply it.
Money, send, delete: your sign-off required. Non-negotiable.
Real incidents studied. Allowlists, audit logs, sandboxes.
🛡 You'll leave more careful than the average AI user, not less.
Ten days, together.
Pods that check your work every day, live build sessions, and a Demo Day stage. This is what the last sprint looked like.
Built for finishers.
Read this honestly before you apply. It saves us both time.
This is for you if…
- You have real work you'd love to delegate: research, content, ops, outreach, reporting
- You can commit 5-7 focused hours on each of the 10 days
- You don't code, and that's fine. The live track is no-code-first; agents do their own setup
- You're technical? Builder extensions go deeper: CrewAI, LangGraph, self-hosting, local MCP
- You'll ship daily and demo in public. The Day 10 capstone is the whole point
This is not for you if…
- You want a course to watch on your own schedule. This is live and intense
- You can only attend some of the days. The ship-every-day chain breaks without you
- You want theory and certificates. The only certificate here is a running workforce
- You won't let agents touch any tools at all. Gated and safe, but hands are the curriculum
Every day ends with something shipped, verified by your pod. Fall behind? Drop to the MVP path, not to nothing.
No private data before the Rule of Two. Human approval on money, send, delete. Never negotiable.
Ask your pod before Google. Verify each other's ships daily. In a room of 15, nobody disappears.
In their own words.
From builders in our previous cohorts.
“I am (98%) hands free on my Dubai business thanks to the skills I've learnt through the course… I'll go from an 8 man team doing 1.5m a year to a 2 man team doing the same without all the staff costs. That's massive ROI on what we paid for the course.”
“Very detailed. On point. Easy to follow. Incredible. For anyone who has decided to launch, this is a must-attend.”
“I built a full website with email integration in 3 hours.”
“I recommend this for anyone with expertise in one field who hasn't had the chance to learn the rest. You leave able to create completely shippable products by yourself.”
“The best course that upgrades your mindset.”
“Support, knowledge, community. Megan is a rockstar.”
“You wish to challenge yourself? The first hackathon gave me an adrenaline rush right to the end.”
One price. Everything in.
- 30 live sessions: 10 days, 3 hands-on sessions a day, taught by building
- Facilitator eyes on your build, daily. 15 spots means nobody's invisible
- A pod of 3-4 that verifies your ship every day
- Agent Mentor, the guardrailed AI coach, whenever you're stuck
- Homework packs plus builder extensions for the technical track
- Capstone, public Demo Day, and alumni repo-watch after
🔑 You bring one tool: a ChatGPT plan with Codex access. Everything else is free or free-tier; expect about $5-15 in API credits across the sprint.
🛡 No refunds. Instead, the Ship Promise: show up, do the work, and we don't stop until your workforce ships.
Join the waitlist for the next cohort
Enter your email and we'll send you a sign-in link. You're on the list, and the first to hear when the 15 spots open.
Everything you'd ask anyway.
What exactly is an “AI workforce”?
Think of ChatGPT as a brilliant intern you have to sit next to all day. An AI agent is that intern with hands: it can use your tools, browse, file, and send, with your approval. An AI workforce is a small team of those agents with different jobs, coordinated, with you as the manager.
I can't code. Can I really do this?
Yes. The entire live track is no-code-first, by design. You build your first agent by clicking, not coding, and modern agents set themselves up: you manage them. Most people who “can't code” have put something live on the internet within three hours of Day 1.
I'm a developer. Will this be too basic?
The live sessions are tool-agnostic on purpose, but every homework pack has builder extensions: rebuild your agent as a CrewAI crew, run a local MCP server, self-host on a VPS, wire Mem0 into your own stack. What most developers actually get is the security discipline, the eval habit, and a finished, demoed system in 10 days.
How much time do I need per day, really?
Three hours of live sessions, max, then 2-4 hours of self-work where your build actually happens. Call it 5-7 focused hours on each of the 10 days. It's a sprint, not a stroll: for ten days, your AI workforce should be your main project.
Is it safe to connect agents to my real accounts?
Only if you do it the way we teach it, which is why security is woven through the sprint, not bolted on. Threat model on Day 1 before any real data. A hard gate on Day 5: no tool touches private data until you can state and apply the Rule of Two. Human approval on money, send, and delete, always.
Tools change monthly. Won't this be outdated soon?
Honest version: yes, the tools will change, and the program is built for exactly that. What you're actually learning is the 5-layer model and the durable capabilities underneath it, the map that has survived every tool cycle so far. Graduates join a monthly repo-watch, so staying current becomes a 30-minute habit.
In ten days, you either have an AI workforce, or you're still chatting.
15 spots per cohort, reviewed personally. Bring your laptop and three workflows you'd love to automate.