Free while in beta
Study anything.
Still know it in six months.
Bring a certification guide, a course reader, a language text or a speech you have to deliver word-for-word. It becomes practice that returns right before you forget, and every card built from your material carries the exact passage it came from.
Cards you generate from a topic instead of a document have no source, and they are labelled that way. That disclosure is the promise: if a card can't cite its source, it says so.
No card required. Paid tiers are planned; see pricing.
- Scheduling
- FSRS-6
- Citations
- Substring-verified
- Grading
- Commit before reveal
- Cost today
- Free in beta
How it works
Four steps, no deck building
The work you would otherwise spend authoring cards goes into practising them instead.
- 01
Add the material
Paste text, upload a document, or point at a web page. Each subject lives in its own space, so terms from one course never get scheduled against definitions from another.
- 02
Cards are generated and checked
The app pulls items out of the material and stores the exact span each one came from. That span is validated as a literal substring of the source before the card is accepted.
- 03
Practise on a schedule
FSRS-6 returns each card shortly before you are predicted to forget it. A fixed daily budget therefore covers far more material than re-reading the same chapter does.
- 04
Track what actually stuck
Mastery needs three consecutive correct recalls, three weeks of retained interval, and at least one confident answer. Confidently wrong answers are surfaced first.
Verification
Where the answer came from, or an honest blank
Generated study material is easy to produce and hard to trust. This is the part that makes it checkable.
- Cited spans are validated as literal substrings of the source chunk before a card ships.
- Ambiguous matches escalate to an entailment check rather than being waved through.
- Stored quotes are re-checked against the live source, and drifted quotes are flagged.
- Topic-generated cards have no source, and the card says so instead of implying one.
What threshold does the standard set for annual review?
“Each control shall be reviewed at least annually, and after any material change to the system it protects.”
Section 4.2, page 18
Practice
Two ways to be tested, one way to be graded
Both modes make you produce the answer first. Recognition is the thing that feels like knowing and is not.
Concept mode
Write what you remember and commit it. Only then does the model answer appear, and the app checks that the anchor term is genuinely in what you wrote before it will accept a correct grade.
Verbatim mode
For anything that has to come out exactly as written: a speech, a formula, a legal definition. The text fades progressively until you can produce it unaided.
Placement check
A short adaptive check estimates what you already hold, per topic, so you start where the gaps are rather than at chapter one. Retakes compare against the previous run.
Use cases
What people bring to it
Each of these is a page, not a bullet, because the workflow genuinely differs.
Studying for a certification exam
Turn the official study guide into scheduled recall practice, with every card traceable to the page it came from.
Read moreMBA and executive education
Keep frameworks, formulas and case vocabulary retrievable across a whole programme, not just until the next exam.
Read moreLearning a language
Build vocabulary and phrase recall from the material you actually read, not a generic frequency list.
Read moreDelivering a speech word-for-word
Rehearse a keynote, toast, script or testimony until you can deliver it unaided, with the text fading as you improve.
Read moreGetting up to speed in a new job
Turn internal handbooks, runbooks and product docs into recall practice so you stop re-reading the same wiki page.
Read more
Ready-made courses
Or start from something already built
Join a catalog course to study the shared version, or fork it into a copy you own and can edit.
- Beginner3 hours
AI & Prompting 101
The working vocabulary and habits behind getting reliable output from a language model, including how to check what it tells you.
- Beginner4 hours
Excel & Spreadsheets Essentials
Lookups, pivots, absolute references and the formula patterns that carry most day-to-day spreadsheet work.
- Advanced5 hours
Advanced AI Practice
Retrieval, evaluation, structured output and the failure modes that show up once a model is doing real work.
- Intermediate10 hours
MBA Core
The core frameworks and formulas from finance, strategy, marketing and operations, kept retrievable rather than skimmed.
- Advanced8 hours
EMBA Leadership & Strategy
Executive-level strategy and leadership material, organised for recall across a whole programme rather than one exam.
- Beginner4 hours
Communication & Public Speaking
Structure, delivery and the verbatim practice that turns a written talk into one you can give without notes.
- Beginner6 hours
Pub Quiz Essentials
The general knowledge a quiz actually tests: science, arts, music, film, sport, myth, language and invention.
- Beginner5 hours
World Geography
Capitals, borders, rivers, oceans, climate zones and map skills, learned so you can recall them rather than recognise them.
- Beginner5 hours
History's Turning Points
Ancient world to the Cold War: the causes, dates and consequences behind the moments that changed direction.
- Beginner5 hours
Science You Half-Remember
Atoms, energy, evolution, genes, plate tectonics and germs, with the common school-science myths dismantled.
- Beginner4 hours
Space and the Universe
The solar system, stars, galaxies, cosmology and spaceflight, taught for recall rather than for awe alone.
- Beginner5 hours
The Human Body Explained
How your body works system by system, from skeleton and heart to gut, hormones and the senses.
Questions
The short answers
- What is this app for?
- It turns material you are accountable for knowing into scheduled recall practice. You create a space for one subject, add the material as pasted text, an uploaded document or a web page, and the app generates cards from it and schedules them. Cards built from a source keep the exact passage they were drawn from, and that quote is checked against the source before the card ships. Practice happens in two modes: concept, where you explain something in your own words, and verbatim, where you reproduce a passage word-for-word. It suits certifications, degree modules, languages, speeches and internal documentation.
- Does every card cite a source?
- No, and the app says so on the card. There are two ways to make cards. If you supply material, the generated card stores the exact span it came from and that span is verified against the source before the card is accepted. If you instead describe a topic and let the app draft cards for it, there is no source document, so those cards are labelled as unsourced. That label is the honest part of the claim: the guarantee is not that every card cites a source, it is that a card which cannot cite one tells you.
- How is a citation verified?
- The cited span is validated as a literal substring of the ingested source chunk before the card is accepted. Exact string matching catches the common failure, which is a quote that was paraphrased or invented during generation. Cases that are ambiguous, for example where whitespace or formatting differs, escalate to an entailment check rather than being waved through. Stored spans are also re-checked against the live source over time, so a quote that no longer matches, because the document changed, is flagged as drifted on the card quality report instead of continuing to look trustworthy.
- How does self-grading avoid cheating itself?
- In concept mode you write your recall and commit it before anything is revealed. The model answer, rationale and anchor term are not in the page at all until you commit, so there is nothing to peek at. After committing, the app checks that the anchor term genuinely appears in what you wrote, and asks a short binary question about it. A blank or off-target answer cannot be marked correct. This matters because generous self-grading is the standard failure of self-directed spaced repetition: the schedule is only as good as the grades feeding it.
- What does it cost?
- It is free during the beta period, with no card required to sign up. Paid tiers are planned, and the terms of service state plainly that accounts may need to move to a paid tier as costs become clear. What stays free after the beta has not been finalised, and the pricing page says that rather than implying unlimited free use indefinitely. Nothing you have already created is taken away when pricing changes; the change would affect what you can add next, and it would be announced before it took effect.
Bring the thing you keep meaning to learn
Free while in beta, no card required. Start with a placement check and you will know within ten minutes where the gaps actually are.