Skip to main content

​
Goal

Return word coordinates so your app can highlight exact words on the source PDF, not just larger text blocks.

​
Sample Document

Use the Bank Statement to test overlays on dense tabular text. Bank statement with word-level bounding boxes overlaid

​
Request

curl -X POST https://api.runpulse.com/extract \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "file_url": "https://platform.runpulse.com/api/examples/637e5678-30b1-45fa-acc4-877f2d636419/pdf",
    "extensions": {
      "alt_outputs": {
        "wlbb": true
      }
    },
    "storage": {"enabled": true}
  }'

​
When To Use It

Use caseWhy word boxes help
Redaction previewHighlight exact words before applying a redaction layer.
Reviewer UIJump from extracted value to exact source words.
OCR QACompare expected values against word coordinates and confidence.
Evidence packsPreserve a visual trail for downstream audits.

​
Checks

  • Word-level boxes are PDF-only.
  • Results are larger; avoid enabling this on every ingestion job.
  • For many products, standard bounding boxes are enough.
  • Store page number, polygon/box coordinates, and extraction ID with review annotations.
  • Test overlays at multiple zoom levels before users rely on them for redaction.

Word-Level Bounding Boxes

Word-level option details.

Bounding Boxes

Standard bounding box format.