Eight students, eight .apkg files, one merged Anki deck that does not break.

The classic recipe is: pick a coordinator, divide the syllabus across members (one or two lectures each), each member makes cards for their assigned lectures inside their own Anki collection, then exports a .apkg per lecture and shares the files. The coordinator imports each .apkg into a master collection and re-exports a combined .apkg, which everyone re-imports. The workflow works on paper. It breaks in practice because eight people writing cards on different note templates produce a master deck where half the cards are missing fields, image-occlusion is inconsistent, and quality varies by an order of magnitude across members.

Anki keys cards on note types (the template that defines fields like Front, Back, Extra). When two members import .apkg files that both define a note type called 'Basic' but with different field orders or different fields, the second import is supposed to be flagged, but a lot of users click through and end up with cards whose fields render in the wrong slots. Multiply that across eight members and the master deck has cards where the answer renders in the source-citation slot, or images render as alt text. Studyly's .apkg ships with a Studyly-namespaced note-type set (studyly_mcq, studyly_cloze, studyly_image_occlusion). Eight members all generating exports use the same three note types, so the merged collection has exactly three note types, not twenty-four.

AnkiHub is real and it is good at what it does, which is community-maintained Step decks with a suggestion-and-approve workflow. The pricing is $6/month Core, $10/month Premium, $450 lifetime, per member. For a 30-person class deck that exists for one semester and then does not, the per-seat cost adds up faster than people expect. The bigger issue is upstream of cost: AnkiHub still expects each contributor to write the cards, so it solves the distribution problem but not the writing problem. Studyly addresses the writing problem (each member gets a 60-second-per-lecture export) and the distribution problem (Anki .apkg files merge cleanly because of the namespaced note types). The two are complements; some classes use both.

If three members each take four lectures and run them through Studyly in their own browser tabs, total wall-clock time is about four minutes (since each 90-slide deck takes 60 seconds and members run in parallel). The longer step is the human review pass per member, which is where the value of having one specific person assigned to one specific lecture pays off: the assigned person spots subject-matter errors that the auto-generation cannot. Realistically, for a 12-lecture block, two members with parallel review get a publishable shared .apkg out the door in 90 minutes including the merge.

Yes, and this is the failure mode that wrecks the most class decks. When members hand-write cards, image-occlusion takes 10 to 15 minutes per labeled diagram, so most members skip it. The merged deck ends up with text-only cards on anatomy, biochem pathways, and histology, which are exactly the subjects where the practical exam is image-based. Studyly extracts the figure off each slide, identifies 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. After merge, the master deck has consistent image-occlusion across every member's contribution, not whichever members felt like doing it.

Anki's import treats every note as identified by a hash of its fields, so two near-identical MCQ cards from different generations import as two notes. You get duplicates. Studyly's review pass before export shows a duplicates warning if the coordinator imports two copies of the same lecture's .apkg into the master collection; the fix is to suspend or delete the older import. The coordination fix is simpler: lock the lecture-to-member mapping in a shared sheet before generation starts.

Edit before merging. Each member runs their lectures through Studyly, opens the deck in the browser, edits stems that need correction (the eval scored 81.3 on a held-out three-document rubric, meaning roughly 4 in 5 cards ship as-is, the rest benefit from a 30-second tweak), then exports the .apkg. Editing inside Studyly is faster than editing inside Anki because the source slide is one click away from every card. The exported .apkg carries the edits.

Add lectures incrementally. Each week, the assigned member generates the new lecture, exports its .apkg, and shares it. The coordinator imports just that .apkg into the master, and re-shares the new master. Anki's filtered-decks feature plus a tag like added:7 lets the coordinator extract just this week's contributions for sanity-checking before broadcasting. You do not need to regenerate the whole deck to add one lecture.

A simple sheet works: Lecture, Assigned member, Generated (Y/N), Reviewed (Y/N), Exported (Y/N), Merged (Y/N). The coordinator is responsible for the merge column. Each row tracks one .apkg from generation through to landing in the master deck. The point of the sheet is not project management, it is so nobody generates the same lecture twice and nobody assumes someone else has already done lecture 7.

Yes. The Studyly free tier on app.jungleai.com covers full lecture generation including .apkg export, no credit card. The honest test for a class is: pick one block (six to ten lectures), assign them across two or three members, run the workflow end-to-end, and see how the master deck feels in week one of review. If the cards play in your existing Anki rhythm, scale up. If they do not, you have lost a Sunday afternoon, not a semester.