You have a PDF. Ninety pages, or three hundred slides, or a scanned chapter your professor uploaded at midnight. You open it, scroll, feel slightly sick, and close it.
The problem isn't discipline. It is that a PDF gives you no way in. A textbook has exercises, a lecture has a lecturer — a PDF has nothing but the expectation that you will somehow absorb it.
Step 1 — Map it before you read it
Spend ten minutes reading only the headings. Write them as a list. You now have the skeleton of the material and, more usefully, a count: eleven sections, not ninety pages.
This changes the problem from “learn this document” into “learn eleven things”, which is the difference between a task you can plan and a task you avoid.
Step 2 — Find what is actually examinable
Not all of it matters equally. In almost every course, the testable material clusters in predictable places:
- Definitions — anything introduced in bold or italics
- Classifications — any list of types, categories or stages
- Causal chains — anything phrased as “leads to”, “results in”, “because”
- Worked examples — if it was worth showing, it is worth asking
- Anything repeated across two sections
Prose that connects these is context. Read it once, don't memorise it.
Step 3 — Turn each item into a question
This is the step that converts reading into studying. For every item you flagged, write a question you could be asked.
Keep them atomic: one fact per question. “List the four stages and explain each” is four questions in a trench coat, and you cannot grade yourself honestly on it.
A rough target: two to four questions per section. For eleven sections that is thirty questions — a real evening's work by hand, and the reason most people stop here.
Step 4 — Answer them before you feel ready
Do not re-read first. Attempt the questions cold, get things wrong, and only then look up what you missed. Getting it wrong before being told is what makes the correction stick — the discomfort is the mechanism, not a sign you started too early.
Step 5 — Space the rest
Tomorrow, redo the questions you got wrong. In three days, redo all of them. In a week, redo the ones that are still shaky.
Three short sessions spread over a week will beat one long session tonight, by a wide margin, for the same total minutes. The reason why is here.
If the PDF is scanned
A scanned document is a stack of images. There is no text to select, which means no search, no copy, and no tool can process it directly.
Photographing the pages with a phone works better than you would expect: on-device text recognition reads printed pages, handwriting and whiteboards. It is often faster than fighting the file.
Doing it in ninety seconds instead
Every step above is mechanical. Finding definitions, writing atomic questions, scheduling reviews — none of it requires judgement, it requires time you would rather spend learning.
That is exactly what Twixem AI automates: import the PDF, get the questions, keep the schedule. The method is the one described here. Only the tedious part is gone.