How the coaching works
That’s not a discipline problem. A one-off quiz has no model of what you’ve mastered, no schedule for bringing it back, and it forgets you the moment the tab closes. MasterTheBook runs on the same class of AI — the difference was never the model, it’s the method wired around it. Every stage of a book gates the answer behind your own attempt first, grades what you actually write, and brings the idea back right before you’d lose it. That’s the one thing an open chat won’t do: make you produce it before it shows you anything.
Don't just read it, Master it.
No card for the free preview. If you subscribe, every tier carries the same 14-day money-back guarantee.
Three moments
It is not one feature you open. It is a sequence of interventions, each placed where memory research says it earns its keep — goal capture and a prediction up front, retrieval-graded study in the middle, spaced review and a demonstrable result at the end.
Before you read
The coach opens by asking what you want out of the book and what you already believe about its central question — then sets a pace-aware plan against that goal. It also asks you to predict how you will do, because writing down a prediction before you start is what exposes the gap between feeling informed and being informed .
While you study
Every chapter ends with retrieval before any explanation or unlock — across 2,260 questions written along the five-stage Transfer Ladder, not generated on the fly. You answer in your own words; the coach grades against the rubric and, when you stall, gives a Socratic hint rather than the answer. During a graded check it goes hint-only — it nudges, but won't reveal — so your score reflects what you know, not what the assistant volunteered. Effortful retrieval is harder than re-reading, and that is exactly why it sticks .
After the session
The coach generates spaced review on the concepts you are weakest on — timed adaptively to when you are about to forget, not on a fixed calendar — and narrates where each idea sits on the ladder. On Master, a spoken oral exam and a graded capstone turn “I mastered this” into something you can demonstrate, because producing knowledge under pressure exposes what quiet recognition hides .
An open chat answers. A coach makes you produce.
You can already ask ChatGPT or Claude to quiz you on a book. The difference is not the model — it is the structure and the memory around it. Six places that shows up:
Chat: Shows you the answer the moment you ask.
MasterTheBook: Gates the answer behind your own retrieval attempt — you produce first, then check.
Chat: Grades you generously, or not at all.
MasterTheBook: Scores your open response against the chapter's rubric and names the exact gap.
Chat: Hands over the solution when you stall.
MasterTheBook: Gives a Socratic hint toward the idea — never the answer — so the effort stays yours.
Chat: Forgets you the moment the tab closes.
MasterTheBook: Keeps a model of every concept on the Transfer Ladder, so it knows what to ask you next week.
Chat: Has no schedule and never checks your confidence.
MasterTheBook: Returns each idea on an adaptive spaced schedule and tracks confidence against result.
Chat: “You've covered it” is a vibe.
MasterTheBook: An oral exam and capstone make mastery a claim you can demonstrate.
AI-native, not bolted on
A chat sidebar is optional by design — you remember to open it, or you don’t, and it has no opinion about when you should. MasterTheBook inverts that. The retrieval gate, the rubric grade, the calibration check, the spaced comeback: each is a research-backed condition the product schedules for you, not a button you have to think to press. The AI is the thing that runs the method, which is why turning it off changes how coaching feels but not what the curriculum is.
Concretely, that means the model is doing real, scoped work — grading an open response against a rubric, generating new practice on a concept you keep missing, narrating where you sit on the Transfer Ladder — always as a scaffold after you have retrieved, never as the first thing you read.
What the AI actually does
No model names, no token math — just the jobs the coach does and the studied reason each one works. Cross-book synthesis is on Scholar and above; the oral exam, capstone, and mastery report are on Master.
It grades what you can produce, not what you recognize
You answer in your own words before any explanation appears, and the coach scores that answer against the chapter's rubric and names the specific gap — the production step Karpicke & Blunt (2011) found outperformed both repeated study and concept mapping. You find out what you can write cold, not what you can pick from a list.
When you stall, it hints — it does not hand you the answer
Get stuck and the coach gives a Socratic nudge toward the idea rather than the solution, so the retrieval stays yours — the kind of desirable difficulty Bjork & Bjork (2011) showed produces better long-term retention even though it feels harder in the moment. An open chat would just print the answer and end the effort.
It writes fresh practice aimed at the concepts you keep missing
When a concept slips, the coach generates new questions on exactly that idea instead of recycling ones you have already memorised — the repeated retrieval Roediger & Karpicke (2006) measured at 61% week-later recall, against 40% for re-reading. You practise the weakness, not the question.
It brings each idea back just before you would forget it
Review is scheduled adaptively against your own performance, surfacing a concept as you are about to lose it rather than on a fixed timer — applying the finding from Cepeda et al. (2008) that the optimal gap scales with how long you need to remember the material. The effort lands where it actually pays.
It keeps an honest read on what you actually know
You predict your performance before each answer, and the coach tracks the gap between confidence and result over time — the correction Koriat & Bjork (2005) identified for learners who routinely overestimate what they know. Familiarity stops passing itself off as mastery.
It connects ideas across books instead of leaving them in silos
ScholarThe coach surfaces where concepts in different books meet and challenges you to reconcile them — turning separate reads into one connected model you can reason across.
It runs an oral exam and grades a capstone, so mastery is demonstrable
MasterA spoken exam and a graded capstone turn “I mastered this book” into a claim you can defend, ending in a mastery report of what you demonstrated under pressure — a study completion record, not a professional credential. Having to produce and structure knowledge rather than review it is what Karpicke & Blunt (2011) found reveals and closes real gaps.
The assessment guard
When you are being assessed, the coach will not reveal the answer before you commit one — it can nudge, but it cannot solve it for you. It also stops re-testing concepts you have already shown you own and that are not yet due for review, so a score reflects what you know, not how generous the assistant was. The measurement and the help are deliberately kept on separate sides of the line.
The non-AI fallback
You can turn AI Coach off entirely. The question bank, the model answers, the adaptive spaced schedule, and the Transfer Ladder are all hand-built and work without spending a single coaching session — you can see how each book is structured on its free preview page, no account needed. The AI speeds up the diagnosis and the feedback; it is the scaffold on top of the method, not the thing being measured. Custom book and PDF upload is coming soon on Scholar and Master — it has not shipped yet.
Decide for yourself
Try Meditations free — create an account, no card. If a week from now you can still explain what you read, the method worked; if it beats your ChatGPT habit, Scholar is $39 a month for unlimited active books and cross-book synthesis — priced next to the other thinking tools you already pay for, not a reading app. Every tier carries the same 14-day money-back guarantee.