Two days, nine pieces, one system running on your data.
You build them in order, each one feeding the next, on your store’s data — every minute of it live. By Sunday evening they converge into one command center that keeps running after you close the laptop.
Your own data. Built live. Yours after the room empties.
One sequence, every cohort, built on your data.
Whatever room you walk into, you build the same nine pieces — each one feeding the next, each one yours to keep. The audience changes; the spine does not.
- 00
Make your data safe to work with.
Before anything touches a model, you learn redact-then-prompt. Strip the PII, keep the signal. Strict org? You run the whole weekend on synthetic data that behaves like yours.
You keep: Redaction playbook + a synthetic-data fallback
- 01
Encode your judgment once.
You write the file that teaches the model how you think — your context, your standards, your stack. Encoded once, it carries into every skill, agent and workflow you build after.
You keep: Your operator-brain context file (CLAUDE.md)
- 02
Turn repeat work into reusable skills.
The tasks you redo every week become named, reusable skills. You stop re-explaining yourself to a chat box and start calling procedures that already know the job.
You keep: A library of skills you re-run on command
- 03
Hand work to agents that finish it.
You build agents that take a goal and complete it — research, draft, reconcile, reply — on your data, with your judgment in the loop. They run while you do other things.
You keep: Working agents scoped to your real tasks
- 04
Give your agents real reach.
Through MCPs, your agents reach the systems you already use — read what they need, act where you allow. Scoped, permissioned, and yours to widen or pull back.
You keep: Scoped MCP connections to your tools
- 05
Wire it to run without you.
You connect the pieces into workflows that fire on a trigger or a clock. The morning report, the follow-up, the sync — handled before you sit down.
You keep: n8n workflows triggered on a schedule
- 06
See your own data answer back.
You stand up a dashboard that pulls from your real sources and refreshes itself. The questions you used to chase by hand now answer themselves on a loop.
You keep: A self-refreshing dashboard on your live data
- 07
Ship a voice and a face on command.
One command takes a brief to a finished video — script, voice, presenter. The update, the explainer, the briefing, produced without a studio or a calendar.
You keep: A Claude to ElevenLabs to HeyGen video pipeline
- 08
Orchestrate it into one operating system.
By Sunday evening the parts converge into one OS — a single place that runs your agents, workflows and dashboard on your live data. You leave with it running.
You keep: Your flagship command-center OS
By Sunday evening, the nine pieces run as one.
The convergence
By Sunday evening the nine pieces stop being nine tools. They wire into one operating system — agents, workflows and a dashboard, orchestrated from a single command center, running on your live data. You leave with it running.
Small, live, built so no one gets stuck.
The whole weekend is live. A lead runs the build up front; two floor coaches work the room, so a stack breaking finds someone fast.
The room
1 lead + 2 floor coaches
A roughly one-to-eight coach ratio so no one gets stuck staring at an error alone.
Seats
25 per cohort
Application-only, one cohort a month. Small enough that the room builds against your stack, not a generic demo.
Code required
None
Natural language end to end. You describe the outcome; the coaches keep you moving.
Pre-work
A short setup pack
You arrive ready — accounts connected, data redacted, environment set. The weekend is for building, not installing.
What “built” honestly means.
You leave with systems running on your own data and a schedule, doing the job by the weekend’s end. Actual output depends on your data and your setup.
Two days gets a working system, not a hardened production deployment with a full team. It runs on your live data and answers real questions; scaling to mission-critical is separate work, which we’re upfront about.
You walk in describing what you wish your tools would do. You walk out having made them do it, on your data, judgment built in, with a runbook to extend it.
Tools change. What you keep doesn’t go stale.
The honest objection to any AI course is that it’s outdated in a few months. We design against exactly that — you own the layer that outlasts the model.
Built model-agnostic.
The orchestration layer is yours. Swap the model underneath — a newer one, a cheaper one, a local one — without rebuilding the system on top.
A runbook, not just a recording.
You leave with the prompts, skills, workflow configs and dashboard templates written down — the steps to rebuild or extend any piece on your own.
Update notes after the weekend.
When a tool in your stack shifts meaningfully, alumni get a short note on what changed. You stay current, and you make the call.
Leave with the system, not the notes.
Every step on this page is on a tutorial somewhere. The gap free content can’t close: your data, a coach watching you build, and a system running when you leave.
vs free tutorials
Free gets you to a plateau.
You can watch every video and still freeze on your own data. Here you build on your data, live, with a coach and a room of peers, and leave with it running.
vs a multi-week cohort
A multi-week program costs you weeks.
Same spine, two days, built in order and finished. You walk out having built the system a longer program only describes across many weeks.
The difference isn’t the information. It’s that the system is running when you leave.
Application-only · One cohort a month · 25 seats
Now find the room built for your week.
Four cohorts run the same spine on four kinds of work — founders, product managers, finance, chiefs of staff. Pick the one that mirrors your week and apply.
Built, or your money back.