Walkthrough · one lecture, .apkg in your collection by the time you finish coffee
The Anki cards your AnKing deck does not cover.
Zanki, AnKing, and Pepper exist because thousands of students iterated on Step content for years. An AI tool will not beat them at boards material this year, and the honest pitch should not pretend otherwise. What those decks do not cover: the specific slide deck your endocrine professor uploaded at 11pm last night, including the labeled diagrams on slides 17, 31, and 44. That is the gap this page is about.
The walkthrough below takes one real lecture (Anatomy I, brachial plexus, 90 slides, 14 labeled figures) from PDF on disk to a deck sitting next to AnKing in your collection. Roughly 90 seconds, no manual carding, image-occlusion included.
Direct answer · verified 2026-05-01
Studyly is the generator. Pair it with your premade deck, do not replace it.
Upload your professor's lecture PDF or PowerPoint, get roughly 200 multiple-choice cards plus one image-occlusion card per labeled diagram in 60 seconds, export as .apkg, import into Anki. Quality scored 81.3 on a held-out three-document eval (factual correctness, clarity, distractor quality, question-type coverage) against Unattle 78.0, Gauntlet 68.0, and Turbolearn 57.8. Methodology on studyly.io/quality.
The .apkg uses Studyly-namespaced note types so importing does not shadow your AnKing or Zanki fields. For Step content, keep your premade deck. For class lectures the premade deck does not cover, this is the fastest path to having cards by morning review.
Why a generator at all, when AnKing exists
The argument against AI flashcard tools in med school is fair: the community decks for boards are better than what any AI will generate this year. The argument for one shows up the moment your class moves off the path AnKing covers. A renal physiology lecture that emphasizes a specific clinical correlation your professor cares about. A pharmacology block where the test is a hundred drug names your professor will pull from her own slide deck, not from First Aid. A gross-anatomy practical where the diagrams on slides 12 through 38 are exactly what shows up tagged on the practical, and AnKing does not have those figures.
That gap was the case for hand-carding for the last decade. An hour or two for 100 cards from a 90-slide deck, then a half hour of image-occlusion if you were rigorous about it. Most students gave up on the image-occlusion step and either re-read the diagrams or hoped the question would not show up. Both strategies lose points.
A class-specific generator with a working image-occlusion path closes that gap. Not for boards, never for boards. For the lecture deck the professor wrote.
One lecture, end to end
Drop the lecture file
PowerPoint, PDF, or scanned image-only handout. The file uploads to your Studyly workspace, gated by the email you signed up with. Nothing public.
Generation runs in roughly 60 seconds
A 90-slide deck produces about 200 multiple-choice cards plus an image-occlusion card per labeled diagram in the source. Distractors are pulled from neighbors of the correct answer, not the open web.
Review the card set inside Studyly
Skim, edit a stem, delete the obvious filler. Editing inside Studyly is faster than editing inside Anki because the source quote is one click away on every card.
Export to .apkg
One click. The package carries multiple-choice, cloze, case-style, and image-occlusion cards with their masks intact. File name matches the lecture title.
Import into Anki, study from your existing collection
File menu, Import. Lands as a top-level deck with the lecture title. Drag it under your class folder. Anki's scheduler takes over from here, alongside whatever AnKing or Zanki cards you already had due today.
What the .apkg actually contains
The package that lands on your disk is a real Anki .apkg readable by Anki desktop and AnkiMobile. The export carries four note types and a source-slide reference on every card. The note types are Studyly-namespaced, which is the only thing that prevents collisions with AnKing's shared field names. If you have ever had a sync go wrong because two add-ons fought over the same field, this is the design choice that keeps that from happening here.
What rides in the .apkg
- Multiple-choice cards with four-option stems, distractors drawn from the lecture's own context (about 200 cards from a 90-slide deck).
- Cloze-deletion cards for high-yield definitions and key terms, useful if your existing Anki routine is cloze-only.
- Image-occlusion cards from any labeled diagram on a slide. Mask sits on the labeled structure, the rest of the figure is visible.
- Case-style stems that build short clinical scenarios from the lecture content (the format that closes the gap to NBME-style class exams).
- Source citation on each card pointing to the slide number in the original PDF, kept inside an Anki field for the explain step.
- Studyly-namespaced note types. Will not shadow AnKing, Zanki, or Pepper note types on import.
Real run on a 90-slide brachial plexus deck. The 14 labeled figures generate 14 image-occlusion cards. Total 218 notes import as a single deck named after the lecture file. Drag under your class folder, done.
Image-occlusion is where the other tools quit
The reason most AI flashcard tools are not useful for med school anatomy is that they drop the figures and write text-only cards. A text-only card asking what attaches to the medial epicondyle of the humerus does not test the same skill as a labeled diagram with the medial epicondyle masked. The labeled-diagram version is what the practical actually tests. The text version is a different question.
Studyly's image-occlusion path takes the figure off the slide, identifies the labeled structures, and writes one image-occlusion note per label with the mask placed over the structure name. The .apkg carries the figure and the mask coordinates. When you study the card in Anki, you see the original diagram with one structure hidden, and you have to recall it. Same skill the practical tests.
For pathology slides, histology micrographs, and biochem pathways the path is identical, since the model treats any labeled figure on a slide the same way. Card quality there is bounded by the quality of the figure on the slide; Studyly does not pull figures from a separate library, so a blurry photocopy stays blurry.
Where the premade deck wins, where this wins
The two are complements, not substitutes. The table below is the version of this comparison I would tell a roommate, not the marketing version.
Use both. Different jobs.
| Feature | AnKing / Zanki / Pepper (community boards deck) | Studyly (your lecture-specific .apkg) |
|---|---|---|
| Coverage of Step 1 / Step 2 high-yield material | Premade decks (Zanki, AnKing, Pepper) own this category. Years of community-vetted card iteration; nothing else is close. | Out of scope on purpose. Use the premade deck for boards content. Studyly is for the lecture material the premade deck does not touch. |
| Coverage of your specific class lecture (next Friday's exam) | Premade decks do not cover one professor's slide deck. You either make cards yourself, an hour or two per lecture, or skip the cards and re-read. | Upload the deck, get ~200 cards in 60 seconds, including image-occlusion cards from the labeled diagrams. |
| Image-occlusion from a slide diagram | Most generic AI flashcard tools drop the figure and lose the fact. Manual image-occlusion in Anki takes 10-15 minutes per labeled diagram. | Extracted automatically. Mask is placed over the labeled structure. Carries through .apkg into Anki as a standard image-occlusion note. |
| Distractor quality on multiple-choice | Generic chat-model output produces distractors that are obviously wrong (filler text, unrelated terms). Held-out eval scored Turbolearn 57.8 on this rubric, Unattle 78.0, Gauntlet 68.0. | Pulled from neighbors of the correct answer in the same lecture context. Held-out three-document eval score: 81.3. |
| Note-type collision with your existing collection | Some tools write cards on shared note types and silently shadow your AnKing fields, which corrupts cards on the next sync. | Studyly-namespaced note types. Imports as a separate deck with separate note types. Removing later is clean. |
| Time from PDF on disk to first review in Anki | Manual carding of a 90-slide deck: realistically 60-120 minutes for ~100 cards, more if you do image-occlusion right. | About 90 seconds end-to-end (60s generation + 30s import). ~200 MCQ + image-occlusion cards. |
Honest limits
A list, since pretending the tool has none is the fastest way to lose a med-student audience.
Cards are bounded by the source. If your professor's slides are thin, the cards will be thin. The generator does not invent content from a textbook to fill the gap; it stays close to the deck so the questions match what the professor will ask. If you need broader context, you upload more sources (a chapter from your textbook, a review article) into the same lecture pack.
Image-occlusion quality depends on the figure. A clean labeled diagram produces a clean masked card. A photocopy of a photocopy produces a card you might want to redo by hand.
Auto-rephrasing on revisit is a Studyly feature, not an Anki feature. When you study the .apkg inside Anki, you study the canonical card set with Anki's scheduler. The wording rotation only happens when you study inside Studyly. Pick the path that matches your existing routine; both work, neither is wrong.
Studyly is cloud-only. If your school requires a fully-offline study tool, this is not it.
Try it on tomorrow's lecture
Drop one PDF. The .apkg is in your downloads in a minute.
Free tier on app.jungleai.com, no credit card. Full export of the first lecture, image-occlusion cards included. Import into your existing Anki collection without touching AnKing.
Common questions about generating Anki cards from your med-school lectures
Should I drop my premade deck and use this instead?
No. Zanki, AnKing, and Pepper exist because thousands of med students iterated on the same Step content for years; an AI generator will not beat that for boards material in the next year. The honest pitch is narrower: keep your premade deck for Step 1 and Step 2, and use Studyly for class-exam content your professor wrote that no community deck covers. The two live as separate decks in your Anki collection. The .apkg uses its own note types, so importing does not overwrite anything in AnKing.
Does the export actually carry image-occlusion cards into Anki?
Yes. When the source slide has a labeled diagram (anatomy figure, biochem pathway, microscopy image, histology slide), Studyly extracts the figure, identifies the labeled structure, and writes an image-occlusion card with the mask placed over that label. The .apkg imports those into Anki as image-occlusion notes. You see the diagram, the structure name is hidden, you have to recall it. A 90-slide anatomy lecture with 12 labeled diagrams produces roughly 200 multiple-choice cards plus 12 image-occlusion cards in one package.
How does this compare to AnkiHub or AnKing's Smart Decks?
Different jobs. AnkiHub is a sync layer for community decks, AnKing's offerings are curated card sets for boards. Both are great for Step. Neither converts the slide deck your endocrine professor uploaded last night into cards by 9am. That is the gap. The product score for question quality on a held-out three-document eval is 81.3 (Studyly) vs 78.0 (Unattle), 68.0 (Gauntlet), and 57.8 (Turbolearn), measured on factual correctness, clarity, distractor quality, and question-type coverage.
What format are the cards generated in, basic or cloze?
Four formats from one source. Multiple-choice with realistic distractors (the default for class exams), cloze deletion for definitions and key terms, case-style stems for clinical reasoning, and image-occlusion for diagrams. The .apkg carries all four. If you only want cloze for your existing routine, filter the imported deck on note type after the import and suspend the rest.
Will importing the .apkg mess up my existing deck or note types?
No. Studyly's .apkg ships with a Studyly-namespaced set of note types, so they do not collide with the AnKing or Zanki note types you already have. The cards land in a new top-level deck with the lecture title; you can drag them into a sub-deck under your class folder, or leave them where they land. If you decide to delete them later, you can remove the deck and the note types together without touching anything else.
Does it work on PowerPoint, PDF, or image-only scans?
All three. PowerPoint and PDF parse natively. Image-only scans (the lecture handout that was photocopied so many times the OCR is hostile) go through OCR first, which adds a few seconds. YouTube lectures work too: the transcript is converted, with timestamps preserved on the explain panel for video sources.
How long does the conversion actually take for a typical lecture?
Sixty seconds for a 90-slide PDF deck on the standard tier. The longest piece is image-occlusion generation when the deck is heavy in labeled anatomy figures, since each figure goes through label extraction. A pure-text physiology deck finishes faster than a 90-slide gross-anatomy deck of similar length.
How does spaced repetition work, since Anki already has its own scheduler?
If you study inside Studyly, the app runs its own spaced repetition with a feature on top: stems get auto-rephrased on revisit so you cannot pattern-match the wording. If you export to .apkg and study inside Anki, you get the canonical card set and Anki's scheduler takes over. The two paths are fine; pick whichever your daily flow already uses. The wording rotation only happens inside Studyly.
Does it handle high-yield identification for pathology and histology?
Yes for any image already on a slide. The image-occlusion path is the same regardless of subject: the figure is extracted, labeled structures are masked, the cards are written. For histology specifically, the cards are only as good as the slide images your professor used; Studyly does not pull from a separate path-image library.
Is the free tier enough to evaluate this for a single lecture?
Yes. The free tier on app.jungleai.com covers a real lecture deck end-to-end, including export, with no credit card. The paid tier raises monthly upload limits and unlocks unlimited generations across decks. The check that actually matters is: import one .apkg, study it for two days, see whether the cards feel like cards you would have made yourself if you had three hours.