Extract structured JSON table rows (headers + rows) and LLM-ready Markdown from PDF invoices, bank statements, receipts, and financial reports using IBM's Docling (DocLayNet & TableFormer) AI models. Built for n8n, Make.com, Zapier, and MCP AI Agents.

• Structured JSON Tables (tables_json): Detects borderless and multi-column tables, exporting clean JSON arrays ready for Google Sheets, Excel, Airtable, or PostgreSQL.
• RAG-Ready Markdown (full_markdown): Preserves headings, reading order, and tables for vector databases (Pinecone, Qdrant, Supabase).
• Zero API Keys Needed: Pre-baked OCR weights run out of the box.

• documentUrls (Array, required): Direct public URLs to PDF documents or invoices.
• extractTablesAsJson (Boolean, default: true): Exports each table as structured JSON rows alongside full Markdown.

{
  "documentUrls": ["[https://arxiv.org/pdf/2408.09869](https://arxiv.org/pdf/2408.09869)"],
  "extractTablesAsJson": true
}