From Prompt to Prototype:
The Cohort
In four weeks you'll ship a working tool that solves a real problem at your job — and you'll know how to build the next one without me.
No payment today. Waitlist gets first access and the founding rate.
The Four P's.
Problem · Prompt · Prototype · Pilot. One buildable cycle, run end to end on a problem from your own job.
Week 1
Problem
Find a real problem worth solving — and define what "solved" means.
Week 2
Prompt
Translate the problem into an implementation-ready spec the AI can build.
Week 3
Prototype
Build, break, diagnose, re-prompt — until you have a deployed, live tool.
Week 4
Pilot
Put it in front of real users and read what actually came back.
Four weeks. One real build.
Each week's output is the next week's input. You leave every session with something concrete.
Problem
Most builds fail because they start with a solution and never find the problem.
- Working Backwards: write the outcome before the product
- Interrogating your own idea — who has this, how often, what do they do today
- Defining "solved" in a way you can actually check
A one-page problem statement, validated against one real person.
Prompt
The translation layer between a problem and a thing that gets built.
- Decomposing a problem into buildable pieces
- What a spec is and why AI needs one
- Writing specs with scope, constraints, and done conditions
- Designing for data capture from the start — what do you need this thing to teach you
An implementation-ready spec, including what your tool will measure.
Prototype
Build, break, diagnose, re-prompt.
- The iteration loop and how to stay out of the doom spiral
- When to fix the prompt vs. when to fix the spec
- Data and persistence — making it hold what users give it
- Deploying and getting a URL you can send someone
A live, deployed tool and a pilot in motion.
Pilot
A prototype nobody uses taught you nothing.
- What a pilot is and isn't
- Designing an experiment: what result would change your mind
- Writing survey and interview questions that don't lead the witness
- Reading what came back — iterate, pivot, or kill
Real usage data and a decision about what's next.
What's expected between sessions
- Before Week 1: bring three things that annoy you at work, plus the name of one other person who has the same annoyance.
- Each week's output is the next week's input — the cohort assumes you did the work between sessions.
- Expect 2–3 hours of build time between sessions.
Demo Day
Each participant walks the cohort through their prototype through the lens of the research they were collecting and through the eyes of their customer.
Not a product demo
A walkthrough of who has the problem, what you built, what you were trying to learn, and what actually came back.
Evidence, not polish
The point isn't a flawless build — it's proof you found a real problem, shipped something, and learned from real users.
Who this is for.
This is for you
Non-technical professionals — ops managers, PMs, marketers, analysts, program leads — who have a workplace problem they'd fix themselves if they could build.
- You have a problem, not a codebase
- You'd build the fix yourself if you could
- You want a repeatable skill, not a one-off
This isn't for you
People who already ship software. If you can build but aren't sure you're building the right thing, that's an advisory conversation — not a cohort.
Book a call insteadThe founding rate.
The first cohort is deliberately underpriced — in exchange for your commitment and your story.
Regular price
$1,900
Founding Cohort · 12 seats
Founding members get the full cohort at the founding rate in return for committing to finish — and allowing their build, demo, and story to be used as a case study. The discount isn't a sale. It's an exchange.
- 4 weekly live sessions (90 min) + Demo Day
- Weekly optional office hours (60 min)
- All sessions recorded
- Single stack, taught deep: Claude + Lovable + Supabase
Founding rate applies to the Fall 2026 cohort only.
No payment today. Waitlist members get first access and the founding rate when enrollment opens.
AJ Bubb
AJ Bubb is the founder of MXP Studio, a fractional AI strategy consultancy specializing in helping mid-market companies navigate AI adoption and digital transformation.
With 20+ years of innovation leadership experience from AWS and Accenture, AJ bridges the gap between technical capabilities and business impact through his Human+AI philosophy.
As host of the Facing Disruption podcast (15,000+ listeners), AJ focuses on practical AI implementation for leaders in healthcare, life sciences, and emerging technology sectors.
Claim a seat.
12 seats, Fall 2026. Waitlist members hear about dates and the founding rate first. No payment today.
This is the most useful thing you can tell us — it shapes the curriculum.
No payment today. We'll only email you about the cohort.