Why SlydeMe exists

Most AI study tools have one fatal habit: they make stuff up. You paste a lecture in, and the AI confidently tells you about cardiac findings the lecturer never mentioned, drug indications that aren't on slide, mnemonics that don't quite fit. It feels comprehensive. It's actually noise.

For board prep, that noise is dangerous. You only have so many study cycles. Drilling a card the AI invented (one that doesn't reflect what your prof emphasized) is worse than not drilling at all. You waste reps and false-confirm gaps you don't actually have.

The strict-scope rule

SlydeMe exists because of one design choice we made on day one: the AI cannot add anything your lecturer didn't say. No related concepts. No "high-yield" filler. No board-style trivia. The output is bounded by the audio and the slides we processed, full stop.

What you study is exactly what was covered. Nothing more, nothing less. That's the whole point.

This makes SlydeMe less impressive on a marketing slide and more useful on exam day. The study guide is shorter than what other tools generate. The Anki gap-fill deck is smaller. The AnKing tag list is more precise. That's because we cut every line that the lecturer didn't actually deliver.

How we enforce it

The synthesis prompt is constrained to refuse outside knowledge. Every claim in the generated study guide has to map back to a transcript timestamp or a slide screenshot. The AnKing matching step is a vector search against an indexed deck, so it does not let the model invent tags. The gap-fill cards are generated only from concepts the matcher flagged as not present in AnKing, then sanity-checked against the source.

It's a small set of rules with a big effect: you walk into your block exam, your shelf, or Step 1 knowing that what you reviewed is what was actually taught. Not what an AI thought sounded important.

That's why SlydeMe exists.