Read incrementally.
Remember for good.
Keep more than you can read in one prioritized queue, extract the useful parts, and turn only the best ideas into memory.
A practice session feels productive when answers come quickly, but fluency during one sitting can be misleading. The stronger signal is whether the idea still returns after time has passed and the context has changed.
Spaced practice deliberately lets a little forgetting happen before the next pass. That gap makes retrieval effortful enough to strengthen the route back to the idea, while still keeping the material close enough that review can repair it.
For a reading system, the point is not to revisit everything on a rigid calendar. High-value extracts should return when another pass can clarify, compress, or turn them into a card; low-value material can wait or disappear from the queue.
Try it: highlight a sentence, then choose Extract.A knowledge refinery, not a read-it-later pile.
Most imported material should fade away. The valuable part moves from source text to extract, then to a small card you can actually keep.
The whole reading loop, end to end.
Import, triage, read, extract, distill, and remember - one local system for people who save more than they can finish.
An extract becomes its own scheduled item while keeping its source, block, and offsets. Cards stay traceable.
FSRS schedules cards for recall. A separate attention scheduler decides when sources and extracts should return.
Stop anywhere. Interleave remembers the exact place, so a hundred parallel sources can stay calm.
Each time an extract returns, trim it, split it, rewrite it, or turn it into a card. Delay becomes a filter.
Command palette, g-letter navigation, and fast card grading. Long sessions stay efficient.
Your library is a SQLite database on your own disk. No account, no cloud sync, no telemetry.
Get Interleave
The macOS build lives on GitHub Releases. Free, open source, and entirely on your machine.