ScribleSketch AI imports for
URL, markdown, and paste.
Ingestion is composition, not conversion. ScribleSketch routes URL, markdown, and paste through the same AI pipeline that powers chat. PDF import is coming soon. The output isn't text. It's a notebook.
4 sources · PDF coming soon · multi-source
Q: crossover? A: meiosis I.
Four ways in. One AI pipeline.
Every source ends in the same place — typed blocks on themed paper. The path each one takes is tuned for the content's shape.
4.2 Cell Division. The process by which a parent cell divides into two daughter cells…
…resulting in two diploid cells genetically identical to the parent. The cell cycle consists of interphase…
PDF import is on the roadmap — column-aware extraction, figure captions, and AI normalization to the block catalog. URL, markdown, and paste work today.
- 1Per-page text extraction (with column awareness)
- 2Image + figure detection (caption pairing)
- 3Heading hierarchy inferred from font weight
- 4AI normalizes to block catalog under target profile
Meiosis is the type of cell division that creates egg and sperm cells, called gametes. Like mitosis, meiosis…
…two stages: meiosis I and meiosis II. Each is similar to mitosis…
URL imports fetch the page, extract main content (favoring article markup), and route through AI normalization.
- 1Fetch via server-side proxy (no CORS gotchas)
- 2Reader-mode extraction (Mozilla Readability)
- 3Image references resolved to inline blocks
- 4Code blocks + tables preserved as typed blocks
Markdown lands with structure intact — heading levels, lists, code blocks, callouts. ScribleSketch-specific blocks lift straight in.
- 1Parsed via remark · structure-preserving
- 2Headings → heading blocks at matching level
- 3Lists / tables / quotes → typed equivalents
- 4ScribleSketch fences (```sketchbook) become real blocks
Q: crossover? A: only in meiosis I — drives variation.
— lab scribbles, pre-class
Paste anything. ScribleSketch detects the format — markdown, code, structured text, free prose — and routes accordingly.
- 1Format auto-detection (heuristic + AI fallback)
- 2Cleanup: smart quotes, line endings, indents
- 3Routed to the matching importer behind the scenes
- 4Result: typed blocks · zero copy-paste rot
Page by page. Block by block.
PDF import is coming soon. This is the pipeline we're building — column-aware extraction, heading inference, figure captions, and AI normalization to the typed block catalog.
5,800 words become 320 — not summarized, but composed. The AI picks which blocks anchor the page; the rest become reference notes attached to those blocks.
Three sources. One thread.
Drop a URL and a paste in one pass — PDF import is coming soon. The AI threads sources into a single composition, and every block carries provenance back to where it came from.
Hover any block to see its source. PDF import is coming soon — URL and paste provenance work today.
Sources are long-lived records.
Imports don't disappear after they're composed. Every source lives in your library — searchable, re-threadable, reusable across new notebooks without re-uploading.
- biology-w4-lecture.pdfComing soonPDF · coming soon · added 2 weeks agoCell biology studyMeiosis primer+3 more×5reused
- khanacademy.org/meiosisURL · article · added 8 days agoMeiosis primerCell prep+1 more×3reused
- lecture-notes-w4.mdMarkdown · 34 § · added 5 days agoWeek-4 reviewFriday quiz×2reused
- lab-scribbles · ⌘VPaste · 847 chars · added yesterdayFriday quiz×1reused
Drop the same source into a new notebook — the AI re-extracts in your new profile's register.
Re-import a newer version of a source; every notebook that uses it gets a refresh prompt.
Click a block, jump to the exact paragraph in the original PDF or URL it came from.