Invoice Data Extraction from PDF

Invoice Data Extraction from PDF

Stop copying invoice data by hand. Provide a PDF URL and the InvoiceExtractor agent calls the LlamaParse tool to get clean markdown, then returns a structured JSON object with every invoice field — ready for QuickBooks, Xero, or a DataTable.

What this workflow does

1. InvoiceURL — provide the PDF URL (paste it in or swap for a Webhook trigger)
2. InvoiceExtractor — agent receives the URL, calls the llamaParseAPI tool, and returns structured JSON
3. llamaParseAPI — HTTP tool node: agent POSTs the PDF URL to LlamaParse and gets back clean markdown
4. InvoiceData — output with structured invoice JSON (vendor, number, date, total, line items)

Use cases

• Automated AP data entry from emailed PDF invoices
• Invoice pre-processing before uploading to accounting software
• Batch ingestion of historical invoices into a DataTable

Setup

1. Open the llamaParseAPI node, replace YOUR_LLAMAPARSE_KEY in the curl command with your real key from cloud.llamaindex.ai.
2. Open InvoiceExtractor and connect an OpenAI-compatible credential.
3. Run once with the sample URL — the agent calls LlamaParse and returns a JSON invoice object in the output panel.

Notes

• LlamaParse handles scanned PDFs with OCR automatically.
• Extend the JSON schema in the agent's system instruction to capture PO numbers, tax IDs, or additional line-item fields.
• For high volume, add a Loop upstream and pass each PDF URL through the same workflow.