The Scheduler: A Study Plan Run by an Algorithm, Not an AI Guess
Your study plan should be maths, not vibes. Given what you know, what is due, and what is coming, it should compute the same principled answer every time, not guess a new one each morning.
The hardest question in studying is not "how do I learn this?" It is "what should I actually do today?" Get that wrong and you either burn hours cramming the wrong things or drift and quietly fall behind. Most study tools answer it in one of two poor ways: they dump every overdue item on you at once, or they hand the question to an AI and let it guess. Norudit's scheduler does neither. It runs a deterministic algorithm to decide what you study, in what order, every day. That word, deterministic, is the whole point of this piece.
Deterministic, not a guess
Deterministic means this: given the same situation, what you know, what is due, and what is coming up, the scheduler produces the same principled plan every time. It is reproducible and it is explainable. The plan follows from your data, not from a model's mood.
Contrast that with an AI planner. Ask a chatbot "what should I revise?" and you are trusting a black box. It can give a different answer tomorrow for no reason, invent a deadline that does not exist, or drift toward whatever it happened to weight most that day. You cannot check its working, and you certainly should not hand it the one resource you never get back: your study time.
A study plan you can trust is one you can check.
Norudit's schedule is built instead on two research-grounded algorithms and your real obligations:
- FSRS-6, the modern spaced-repetition algorithm that also powers Anki. It models memory and predicts when you are about to forget something.
- Bayesian Knowledge Tracing (BKT), a model from decades of intelligent-tutoring research that estimates how well you actually know each concept from the way you answer.
- Your fixed obligations: the exams on your timetable, your coursework deadlines, the classes you really have this week.
There is no "trust the AI" step anywhere in it. The plan is computed. That is what lets it be not merely plausible, but genuinely close to optimal for how memory works.
How the priority works
Every day the scheduler weighs four inputs and turns them into one ordered plan.
1. What is due, from FSRS-6. Every concept you have studied has a predicted moment when it is about to slip out of memory. FSRS-6 tracks that for each one and aims the next review to land just before it fades. Review too early and you waste the repetition; leave it too late and you have already forgotten and must relearn. The scheduler surfaces what is due, so you catch things at the most efficient possible moment.
2. What is weak, from BKT. Bayesian Knowledge Tracing estimates, from your answers, how firmly you know each concept, and it updates that estimate every single time you are tested. Weak concepts get pulled forward; the ones you have clearly mastered step back. Your effort flows to the gap instead of spreading evenly over things you already know.
3. What is fixed, from your obligations. Exams and deadlines are hard anchors. The scheduler works backward from them, so an exam two weeks out changes what lands on today. Nothing important is left to the last night by accident.
4. What fits your day. The plan is fitted around your real timetable and life, so it never hands you eight hours of work on a day you have one hour free.
It then combines these into a single ranked list: the highest-leverage things to do right now, given everything at once. Not a wall of cards. Not a vibe. A defensible order you could, if you wanted, trace back to the reason for every item.
A backlog that does not become a cliff
Everyone misses days. Most spaced-repetition tools punish you for it. Skip a week and you return to a wall of hundreds of overdue cards, which is so demoralising that people simply quit. The maths was never the problem there; nobody managed the fallout.
Norudit manages the backlog rather than dumping it. When you miss sessions, it does not pile every overdue item onto one brutal day. It re-solves the plan, spreads the catch-up sensibly, and puts the most important and most at-risk items first. Falling behind becomes something you recover from over a few days, not a cliff you drop off. And because the backlog is recomputed rather than guessed, the plan you come back to is still the right one.
Exam prep, aimed at the real thing
The scheduler knows your exam dates, so it does not just tick over with maintenance review. As an exam approaches, it bends toward it. It can build a timed mock paper aimed at your weakest topics and tie it to the exam on your timetable, so your practice sharpens toward the actual assessment at the right moment rather than at random. The closer the exam, the more the plan is shaped by it.
Reminders, so the plan reaches you
A plan you have to remember to check is a plan you will drop. Norudit emails you your schedule: a short daily reminder of exactly what to do today, so the decision is already made before you have opened your laptop. You can have it email your plan and nudge you each morning. The friction of "what was I meant to be doing again?" simply disappears.
It recalibrates the moment anything moves
Add a deadline, miss a day, finish something early, or have an exam moved, and the whole plan re-solves at once. You are never following a stale schedule. Because it is an algorithm and not a fresh guess each time, that recalculation is instant and consistent. Move one thing and everything downstream quietly rearranges to stay optimal.
The quiet engine
Scheduling is the least glamorous part of studying and the first thing almost everyone gets wrong. It is also the engine that keeps the whole method turning: the right review at the right moment, the right practice before the right exam, and a sane way back when life gets in the way. Norudit gets it right for the same reason you can trust it. It refuses to guess.