Every family we work with has the same story. The student uses ChatGPT for homework. The homework grades look fine. The test scores don't. The parents ask us: do we ban it? Do we ignore it? What's the actual answer here?
We've been coaching families through this question for the last year, one at a time. Enough of them found the same framework useful that we finally productized it.
Today we're launching the AI Study Skills Mini-Course. Two coached sessions, a written personal protocol, and a follow-up check-in a month later. $249 flat. It's built specifically for 6-12 students (grades 7-12) who are already using AI heavily to study and want to keep using it without losing the outcomes they actually care about.
The evidence we couldn't ignore
Two studies changed our framing this year:
- A June 2026 study of 26,811 secondary students found that AI adoption raised homework scores by 18% and dropped exam scores by 20% within six months. Entrance-exam scores fell 18–24% over two years. STEM subjects were hit hardest.
- A Wharton experiment found that students given ChatGPT solved 48% more practice problems correctly, then scored 17% lower than AI-free peers on the actual test.
The pattern shows up for coding students the same way. AI-assisted assignments look strong until the AP CSA mock exam, USACO round, or college interview — where the reasoning has to come from the student.
The temptation for parents (and for adult learners themselves) is to ban AI. The problem is bans don't work — every study of tech restriction in education shows compliance drops within 3–6 weeks. What does work is a framework that specifies when AI helps and when it hurts, applied to the learner's actual work.
The framework: the four phases of learning
Real learning happens in four phases:
Phase 1 — First encounter. You see a new concept for the first time. Light AI use is fine here: ask for a plain explanation or a summary. Keep it short.
Phase 2 — Initial struggle. You try to apply it. You get stuck. This feels bad. AI destroys this phase reliably — because cognitive struggle is what builds the neural pathways a test requires. Every time AI produces the answer here, the brain doesn't encode the reasoning.
Phase 3 — Solve. You produce an answer. Write it down. Explain it. This is the skill your exam or interview will test. Practice it exactly as it will happen: without AI.
Phase 4 — Review. You check your work against the "official" solution. Learn from mistakes. AI is genuinely great here. Use it freely.
The pattern most learners fall into: collapse Phase 1 into Phase 4, skip Phase 2 and 3 entirely. They read the problem, ask AI to solve it, read the solution, and say "yeah that makes sense." Then on the test they're alone with the problem and the reasoning isn't there.
Naming the phase out loud is the first step to catching yourself.
What's actually in the mini-course
Two 60-minute sessions.
Session 1 — Diagnose & Decide. We walk through the learner's actual current habits. Where does AI show up in their study sessions or work? Which phase are they in? We introduce the framework, review the evidence, and draft a first version of a personal rule set — three rules, calibrated to their subject or role. They leave with a 48-hour homework log to run before Session 2.
Session 2 — Rehearse & Ratify. We review the log together. Which rules held? Which slipped? We test the rules against a real problem live, with a coach watching for near-violations. We refine the rules based on real-world friction. For families, the last 20 minutes are a parent-and-student conversation so both parties own the protocol.
Within 24 hours of Session 2, the learner receives a written 2–3 page Personal AI-Use Protocol — their rules, their anti-patterns, their warning signs, their success signal. A follow-up check-in email lands 4 weeks later to see what stuck.
Who this is for
- Parents of grade 7–12 students who use AI for homework and are seeing test scores drop. Rules the student helps draft get followed 3× more than rules a parent imposes.
- High schoolers preparing for AP CSA, USACO, or college app interviews whose AI-assisted coding is about to hit its first proctored test.
- Any 6-12 student in a CS class where AI use is quietly eroding foundational skill-building.
Not for: kids under 12 (too young for the meta-thinking), students already in a serious grade crisis (start with our Report-Card Rescue package first), or anyone who wants a ban rather than a framework.
What it costs, what happens next
$249 flat. Refundable within 48 hours of Session 1 if it isn't right. Add-on to any existing coaching package for a $50 discount.
If you want to book: send a note through our contact form mentioning the mini-course. We'll reply within one business day to schedule Session 1.
If you want to see the full framework first: the landing page has the four-phase diagram and the objection-handling FAQ.
If you want to know if it's right for your student or your career switch: a $99 Diagnostic Session is often the first thing we recommend — the mini-course is a frequent follow-up when a diagnostic reveals AI-driven skill gaps.
Want a plan tailored to your student?
Every engagement starts with a $99 diagnostic session and a written 2-page gap report. Refundable within 48 hours if it isn't a fit.
Book Your Diagnostic →