E-invoice data extraction

GST e-invoices are built for machines: every field on them, the IRN, the supplier and customer GSTINs, the HSN/SAC codes, the taxable value, the tax split, exists in a defined structure designed to be read electronically. But the version that reaches you is a PDF, and the data is still stuck inside it. Dynamite Docs extracts e-invoice data, pulling the IRN, supplier and customer GSTINs, invoice number and date, HSN/SAC, line items, taxable value, the CGST/SGST/IGST split, and totals into confidence-scored rows you validate, match against vendor records, and export for your books.

Try this workflow free or view pricing.

E-invoices are machine-readable, but you still get a PDF

An e-invoice is identified by its IRN, the invoice reference number that ties the document to the electronic record, and the IRN has to be captured alongside the visible fields if the invoice is going to be matched and referenced later.
The data lives in the signed JSON and the QR on the PDF, but what lands in your inbox is the rendered document. Reading the fields off a QR by eye, or rekeying them, is exactly the manual step e-invoicing was supposed to remove.
Validation is the point of e-invoicing. The extracted fields have to match your vendor records, GSTIN, taxable value, tax split, and total, and mismatches have to surface before the invoice reaches the books.
An e-invoice does not sit alone. It travels with an e-way bill when goods move, and it references the same transaction your purchase records and vendor statements do, so the IRN has to tie all of it together.

How Dynamite Docs handles e-invoices

01 — Drop the e-invoice PDFs in

Upload e-invoice PDFs, signed JSON files, or QR screenshots as they arrive from suppliers. Digital e-invoice PDFs are parsed deterministically from their text layer: no AI call, identical results every time.

02 — Extract the invoice fields and the IRN

IRN, supplier and customer GSTINs, invoice number and date, HSN/SAC, line items, taxable value, CGST, SGST, IGST, and totals are read from each layout on the spot. The IRN is captured as a field of its own, not buried in a page corner.

03 — Validate and match

Rows carry the fields that have to agree with your records, so the e-invoice can be checked against vendor records and the purchase side of the transaction. Confidence scores flag the fields a human should confirm, and corrections become patterns the next batch reuses.

04 — Export for the books

Export to Excel, CSV, or JSON, push rows to Google Sheets, or prepare them for entry into Tally. Batch processing handles a month of e-invoices at once.

Document types this workflow handles

GST e-invoice PDFs — IRN-verified electronic invoices parsed deterministically
Signed JSON / QR screenshots — embedded e-invoice data captured along with visible fields
E-way bills — goods-movement documents for the same transaction
Vendor statements / reconciliation files — for matching against e-invoice rows

What gets extracted from a GST e-invoice

The exact fields are inferred per layout. A typical e-invoice run yields these, each confidence-scored:

IRN — 9f8b2c1a6e4d7f3a8b2c5d1e9f4a7b2c8d1e3f5a
Supplier GSTIN — 27AABCS1429B1ZP
Customer GSTIN — 29AABCC3456C1ZD
Invoice number — INV/24-25/0144
Invoice date — 2026-08-02
HSN / SAC code — 4820
Taxable value — ₹24,000.00
CGST / SGST amount — ₹2,160.00 / ₹2,160.00
IGST amount — ₹0.00
Invoice total (incl. tax) — ₹28,320.00

A realistic example

E-invoices extract to rows like these, IRN and tax split captured, ready to validate against vendor records and export for the books:

IRN (first 16)	Invoice #	Date	Supplier GSTIN	HSN/SAC	Item	Taxable value	CGST	SGST	IGST	Total	Status
9f8b2c1a6e4d7f3a	INV/24-25/0144	2026-08-02	27AABCS1429B1ZP	4820	A4 copier paper	₹24,000	₹2,160	₹2,160	—	₹28,320	Matched
7c1d9f4e2b8a3c5d	INV/24-25/0145	2026-08-03	33AABCC3456C1ZD	8467	Rotary hammer	₹1,86,000	—	—	₹33,480	₹2,19,480	Matched
—	BILL-118	2026-08-05	29AABCC3456C1ZD	4820	Desk stationery	₹4,500	₹405	₹405	—	₹5,310	No IRN — flagged for review

The documents behind GST e-invoicing

An e-invoice is a GST invoice verified through the e-invoice system and identified by its IRN, and it is not the same as an e-way bill, which is the document for movement of goods. Keeping the two straight matters in practice: an invoice is matched on its IRN and fields, a shipment on its e-way bill number, and the transaction pulls both together.

The IRN is the field worth treating as a reference, not a detail. It ties the invoice you hold to the electronic record, and it is what lets the same transaction appear consistently in your purchase records, your supplier’s record, and the transport document. Capturing it accurately on every e-invoice is the difference between records that reconcile and records that quietly diverge.

Validation is where e-invoice extraction pays for itself. The supplier GSTIN, taxable value, tax split, and total on the e-invoice have to agree with your vendor record and the bill you already captured. Extracting both sides into rows makes the comparison fast, and confidence scores flag the fields, a handwritten GSTIN, a scanned page, a blurred total, that a human should confirm.

Not every supplier is on e-invoicing. Thresholds mean some vendors still issue ordinary bills, so a real accounts-payable run is a mix of IRN-carrying e-invoices and legacy supplier bills. One workspace that extracts both, with the IRN captured when present and flagged when absent, keeps the whole run consistent instead of splitting it across two systems.

Why bring-your-own-AI matters for e-invoice data extraction

The IRN is the field worth treating as a reference: it ties the invoice you hold to the electronic record, and it has to be captured accurately on every document. The reader that does the capturing is your choice, bring your own key and pay your provider’s rate, with unlimited processing on Pro and Ultra.

Digital e-invoice PDFs carry a text layer and parse deterministically for free, identical results every run. QR screenshots and scanned supplier records are read by the vision model you pick, and where a QR is present the visible GSTIN is cross-checked against it.

Rows carry the fields that must agree, supplier GSTIN, taxable value, tax split, and total, so validation against vendor records and previously captured bills happens on structured data. Corrections compound per supplier, and the exported rows stay yours for the books.

Questions, answered

What is the IRN and why does it matter?

The IRN, or invoice reference number, is the unique reference that identifies an e-invoice and ties it to the electronic record generated through the e-invoice system. Dynamite Docs captures it as a field, so the invoice can be matched and referenced against your records and transport documents.

Can it read e-invoice PDFs where the data sits in the QR code?

Digital e-invoice PDFs are parsed deterministically from their text layer, and QR screenshots or signed JSON files are read by the vision model you choose. The IRN and invoice fields are extracted in both cases.

Does it validate e-invoice data against vendor records?

It extracts the fields that must agree, supplier GSTIN, taxable value, tax split, and total, into confidence-scored rows, so validation against your vendor records and previously captured bills happens on structured data. The validation process itself remains yours.

Is e-invoice extraction the same as e-way bill extraction?

They are separate documents, e-invoices are IRN-verified invoices and e-way bills are goods-movement documents, and Dynamite Docs handles both. Each is extracted into its own structured rows, so the same transaction can be matched across both.