Your notes become your question bank.
Tag facts in your own Obsidian notes and Bottsy turns them into Bitts — exam questions and flashcards — then schedules your review with spaced repetition. Local-first, and read-only against your vault.
From note to question bank
Your vault is the source of truth. Bottsy works from it in four steps — and never writes to it.
Tag a fact
In Obsidian, add a small marker to any note — flashcard, cloze, exam question or diagram. Your notes stay exactly as you wrote them.
Review what sync finds
A sync check proposes new Bitts from what you tagged, each with an advisory AI quality note. Nothing is added until you approve it, and a safety copy of the database is taken first.
Study what is due
Press Study Now. The SM-2 scheduler serves whatever is due, and new Bitts graduate through short learning steps. Answer by clicking, typing or speaking.
Keep the bank honest
A mechanical audit, AI quality sampling and advisory quality reviews catch weak or dated material. Flag a Bitt and the audit can learn a new rule from it.
One piece of study material: a single exam question or flashcard. Every Bitt traces back to where it came from.
A combination of notes and tags you revise as one — an exam block, say — with its own confidence and coverage figures.
What Bottsy does
A question bank, a flashcard system and a scheduler, joined into one local pipeline.
Your vault, mirrored read-only
The Library shows every note, its tags and the Bitts it has produced, with your own formatting carried over. Bottsy reads your vault; it never writes to it.
Your words, not a model’s
Flashcards, cloze and diagram Bitts are derived mechanically from the words you wrote — no LLM rewriting them. Vault exam questions are hand-approved and inserted verbatim.
Questions where no bank exists
Where you want more, Bottsy generates single-best-answer questions in the style of MRCPsych and MRCP papers — plausible distractors, first-principles explanations — using your own Anthropic key.
Spaced repetition, no black box
Standard SM-2 with two learning steps: Bitts you struggle with return within minutes, ones you know at expanding intervals. No streaks, no badges, no proprietary scheduling.
Study a Sett as one
Group notes and tags into a named Sett and study it as a unit, with confidence and coverage figures built from your actual review history rather than self-rating.
Voice study, strictly optional
Speak your answer; Whisper transcribes it and Claude grades it. Questions can be read aloud too. Off by default, with a full text-only path everywhere.
Kept honest by design
Generated content can be wrong, and even your own cards age. Bottsy treats checking as a first-class feature, not an afterthought.
Mechanical audit
Deterministic checks on every Bitt — option counts, duplicate distractors, multi-fact cards, missing explanations. Free and instant.
AI quality sampling
A rolling sample of recent Bitts is reviewed for what rules cannot catch: implausible distractors, giveaway stems, explanations that restate the answer.
Advisory quality reviews
Each Bitt proposed from your vault notes receives an advisory AI quality review before approval, flagging potential issues like weak distractors or multi-fact cards.
Learned rules
Flagging a Bitt can teach the mechanical audit a new rule, so the same problem is caught automatically from then on.
Made by a doctor who studies this way
Active recall and spaced repetition have a stronger evidence base than almost any other way to study. The limitation has never been the method — it is that question banks only cover what someone else chose to write, for the audiences they chose to serve.
Bottsy is built by a UK psychiatry doctor for medical postgraduate exams, working from the notes he already keeps. It covers the exam material — and everything the banks never reach: the specialty interest that came from a paper, the guideline worth understanding from first principles, the research area worth learning properly.
It is a personal tool first, in daily use by the person making it. If the way it works matches the way you study, register your interest below.
Local-first, by design
Your study data does not leave your machine.
What this means in practice
- Everything lives on your own computer: Bitts, study history, settings and costs in a local database
- Your notes stay in your vault — Bottsy reads them and never writes to them
- No accounts, no registration, no cloud sync
- No telemetry, no analytics, no tracking of any kind
- You bring your own API keys (Anthropic, OpenAI, ElevenLabs), encrypted at rest and sent only to their own provider
- Every model call is logged with its cost, and deletion is always reversible
Bottsy is in active development.
If you’d like to hear when there’s something you can try, send a message and we’ll keep you posted.
Register interest