Question tagging keeps running on its own schedule and does not take time from the site work. The bank is at 12,480 questions. 11,632 have passed the fact check, 640 have not been reached yet and 208 are flagged for a closer look rather than quietly passed.
12,050 have had every answer choice checked rather than just the right one, which is the slower half of the job and the part that matters for a practice question. Everything that finishes appears on the question page as it lands. Nothing that has not finished goes anywhere.
Running against the question bank on its own schedule and costing very little of my time, so it continues in the background without competing with anything else here. There is no weekly count to promise; the total climbs steadily.
On the record in one place rather than spread across messages.
The bank is built on ARDMS and sonography material rather than derived from a particular set of textbooks.
Only the questions that have finished the passes are verified. The second pass checks each one and substantiates what it claims against authoritative references; 11,632 of the 12,480 have passed it. Until a question clears it, nothing from the bank goes to you, into a product, or in front of a student.
You found this yourself in June, on the physics interactive, where the question was wrong. A study tool that teaches something incorrect is worse than one that does not exist, because the student trusts it and carries the error into an exam. That is why the pass is slow.
Where the checking looks things up it is held to ARDMS, SDMS, AIUM, ARRT, the ACR, Radiopaedia and NCBI StatPearls. Reddit, allnurses, Quizlet and Brainscape are blocked outright: student flashcards repeat each other's mistakes, and that is how a wrong answer becomes consensus.
You offered PDFs of the books your students are tested from. That is worth more than any reference I would pick, because you know what is examined. If you still want to send them, they go straight in.
You ranked this first of the three, and it is the one that moved furthest.
It searches four places per topic, keeps every image with the page it came from, reads the license on that page, and records what it permits and what credit line is required. Anything whose permission cannot be demonstrated is held back.
From here it stops being something I run for you. The tools go onto your own machine with instructions written for your computer rather than a developer's, so finding images stops being a thing you wait on me for. That is the item you named as the most time consuming part of your own work.
This finds licensed images from outside your material to sit alongside questions. It is a different job from the product preview images on the product images page, which come only from pages you wrote. They never share an image.
The only automation left that needs a decision from you.
The same pipeline as question tagging, applied to written study material. Once it runs, every piece of content can be found by topic and new content matched to what exists.
You have said the site is where you want the focus, so automation gets a fixed, limited number of hours each week. The reason for keeping it moving at all is processing time: work started this month finishes far sooner than the same work started after the site is done.
Drag to reorder, or use the arrow keys once an item is selected. Question tagging and image research are not in the list: the first is running and the second is built and coming to you.