I read the full corpus: positioning, both curricula, session guides, the partner playbook, the lessons and metrics files. The product is strong and the GitHub-native delivery solves problems most training programs never even name. The learning science is thin. The core loop is teach, copy-paste, verify, with no stated success criteria and differentiation only at intake.
That gap is the opening. Instructional design is my day job, and I ship software with the exact tools this program teaches. The cleanest place to start is a vertical the catalog does not have yet: AI for educators, with me as the case study.
The funnel already exists: content on top, courses and cohorts in the middle, the skills marketplace underneath. Two pieces are missing. Video is one; the content engine is text only, and Jeremy asked me on our June 7 call to make YouTube videos. The vertical is the other.
Teachers are the best version of the non-technical founder this program already serves: non-technical, and drowning in work AI can already do. The three-tier shape maps onto teacher professional development (PD) almost one to one. Schools already buy training in exactly these shapes; it is a budget line every year.
The Session 1 guide is honest about its method: the student copy-pastes the full CLAUDE.md block from the handout. Fine as a we-do. The problem is what never comes after. At no point does the learner build something unaided and get measured against stated criteria.
Commits get celebrated at five, ten, twenty-five. Commits measure activity, not learning. A student can commit twenty-five times and still be unable to start a project alone.
Start from what the learner can do when the session ends, design the check that proves it, then the instruction. No session in either curriculum names its exit skill. The fix is small: a success criterion per session, a formative check mid-session, a completion rubric per tier.
In my Oregon district I worked on power standards: out of the full pile of state standards, name the small set that carries the weight, write proficiency scales for each, hang everything off that spine. The positioning doc says credentials should mean something. A spine is how a credential gets to mean something. Without one, CCA is a certificate of attendance.
Today the model is Jeremy teaching one student at a time. The session guides are already templates, the right bones. Missing: scripted decision points, misconception banks, fidelity checklists, an observation rubric. That layer is what lets the program scale past one calendar.
One principle runs through everything I build. In schools, the behavior system (PBIS) counts behavior compliance and the gradebook counts academic compliance; agency is the quadrant neither tool tracks, and PANDA, the classroom platform I built and teach with, reads it. The rebuilt module gives adults the same respect: milestones framed around autonomy, no gold stars. Motivation design that fits the learner's age.
The corpus teaches the I-do and the we-do. The you-do never arrives.
I own this. The three tiers rebuilt for a teacher buyer: a rubric generator, a parent-email drafter, differentiated text sets, a grade-export analyzer. A fifth Education track lands in the template repo, and the team tier becomes a department cohort. My classroom is the lab; that is how PANDA got built.
Success criteria, you-do checkpoints, formative checks, completion rubrics, a competency map for the 32-week program. It changes nothing about what Jeremy teaches; it changes what the learner can prove afterward. The proof exists: Schema Studio rebuilds the 60-minute database session with the full layer, original and upgrade side by side. Session 1 stays the diagnosis; the database session is the rebuild, because it carries a transferable skill worth assessing.
The case study is me. A middle school ELA teacher with no computer science background who shipped, on nights and weekends, a live finance dashboard a private investor uses daily, a prediction-markets bot in production, and PANDA, the classroom app my own students use. Same tools, same loop this program teaches. Every teacher watching can see themselves in the starting point.
Two proofs are already live for this plan specifically: the PANDA demo, synthetic students, nothing to install, and Schema Studio, the rebuilt Intent Solutions module described in the roadmap. Neither is a mockup.
The build log. A teacher shipping real software in public.
The searchable tutorial: plan a unit with Claude, differentiate a text set, draft the IEP notes. The person searching already has the problem.
Every video repurposes through the engine Jeremy already runs: LinkedIn and X posts, a field-notes entry, cross-posts with canonical links. One production, four surfaces.
Three milestones across four to eight weeks, one acceptance criterion each, an AAR and a metrics file at close. The same discipline the partner playbook applies to everyone else.
A written pedagogy audit of both curricula. Accepted when Jeremy signs off on the audit and its prioritized fix list.
Education track v1, Schema Studio polished and deployed, competency spine v1. Accepted when a teacher who is not me completes the rebuilt session and hits its stated success criteria.
Pilot YouTube series live, education track listed on the marketplace. Accepted when published and listed.
The validated ladder, rebuilt for teachers. Starter, build-with-me, and a team of five as a department.
Five to twenty-five teachers, flat fee, a shape school budgets can approve.
Market norm for outside providers, labeled as such, not a tested number. Train-the-trainer licensing comes later, and it is where the long-term money is: selling the program instead of hours.
Specific splits and figures stay out of public artifacts, same as the rest of the corpus. This section prices the work. The split depends on the structure below.
Three alternatives. Each one works on its own.
Head of education. The vertical is mine to run: curriculum, the teacher track, the YouTube lane, train-the-trainer. Compensation is a revenue share on the vertical. Fits if Jeremy wants the arm to exist and does not want to run it.
A stake in Intent Solutions or a jointly owned education entity. Fits if we both read this as a long bet. The heaviest option, and the one that most needs paper.
Run M1 through M3 at a defined fee, then decide. Lowest commitment, prices the work honestly, and exactly what the playbook says to do with a new engagement type: run it once, write the AAR, then decide.
The instructional-design layer is the part of the catalog that does not exist yet, and it makes everything else more sellable. I am an instructional designer who is also the case study, with a classroom to test in and shipped software to point at.
Teaching is the day job and the build window is nights and weekends. I am the person proposing to add YouTube production to a schedule that already includes middle schoolers. Scope is the fix: two videos a month, and the audit lands before any filming.
Both consulting sites are quiet today, and YouTube is unproven for us both. The tutorial lane mitigates: searchable content compounds while subscriber counts are small.
Jeremy is, by his own description, in an exploratory phase. That makes starting easy and finishing hard. The three-milestone window exists precisely to bound it.
District procurement runs months. The teacher-direct tier sells to individuals while the pipeline warms. Separately, PANDA is back in service as the proof asset; the open cost there is maintenance hours, not direction.
Pedagogy-layer ownership must be settled before the upgrade ships, under any structure. And the channel question: our June 7 understanding was that I can take my own brand and go if it takes off. That should be written down while we are friends, because that is when these things get written down well.
M1 audit delivered. Schema Studio and the PANDA demo polished and live as the public proof; both already exist as working builds. Structure agreed. First two pilot videos shot.
Education track v1 lands in the template repo. The teacher-edition tier sheet is final. Four to six videos live across both lanes. First paid teacher cohort or pilot PD session booked, even if small.
M3 closed: series live, track listed, case study on the portal. Competency spine v1 across the 32-week program. Then the decision point, made from the metrics file rather than from momentum: double down, restructure, or stop cleanly.