Splitters detect logical boundaries so you can process long documents in parallel.

Create a splitter

1. In the Studio, go to Processors → New → Document Splitting.
2. Choose a strategy:
   • Layout-based for PDFs with clear page breaks.
   • Semantic for text-heavy documents using embeddings.
3. Upload at least three examples so the model understands typical structure.

Rules & tags

• Add page range rules to force splits (e.g., every 10 pages for statements).
• Define section tags like introduction, appendix, or invoice to enrich outputs.
• Combine with workflows to trigger different extractors per section.

API deployment

curl -X POST https://api.algorythmos.fr/processors \
 -H "x-api-key: $ALG_KEY" \
 -H "Content-Type: application/json" \
 -d '{
  "type":"splitter",
  "name":"policy-splitter",
  "config":{"strategy":"semantic","min_confidence":0.8}
}'

Publish the splitter and wire it into your workflow to distribute downstream processors automatically.