Extract GST invoice fields and line items to JSON using AI

Header fields, nested items Illustrative workflow.

By Dynamite Docs. Published 2026-08-14. Updated 2026-09-30.

Turn GST invoices into clean, reviewed data for Excel and Tally.

GST invoice to JSON conversion is a mapping job, not a text dump. Dynamite Docs proposes document fields and repeating line items from digital PDFs, scans or photos, then lets a reviewer check GSTIN roles, references, tax values and totals before export. The product JSON preserves columns, rows and document metadata so developers can map the reviewed result into their own contract. Use the workspace for manual review, or connect paid-plan API and webhook features when an application needs a repeatable handoff.

Scope and tax review

Content owner: Dynamite Docs content team. Official-source review last updated: 2026-09-30.

Dynamite Docs extracts fields from source documents. It does not determine whether an invoice complies with GST law, prepare a return, or make a document ready to file. Review every extracted value against the original and use the current GST Portal guidance or a qualified tax professional before filing.

Choose file storage and an AI provider that meet your document requirements. Review encryption, retention, own-key processing, and local Ollama options in our Security & Compliance controls.

Official GST sources

CGST Rules: tax invoice particulars (Rule 46), Central Board of Indirect Taxes and Customs. Mandatory 16 invoice particulars and tax invoice format requirements.
Form GSTR-1 preparation and Returns Offline Tool FAQs, GST Portal. GSTR-1 invoice data, preparation modes, validation, and error correction.
Form GSTR-2B viewing and reconciliation FAQs, GST Portal. Auto-drafted ITC statement, downloads, cutoff dates, and book reconciliation.

Try a GST invoice or compare its storage, processing, and export limits.

GST INVOICE DATA: Send invoices via UI or API → Standardized field extraction → Automated webhook delivery. Illustrative workflow.

Software needs structured JSON, not PDF documents

Accounting platforms and custom software need a versioned contract, while an extracted invoice can introduce fields or labels that vary by layout.
Different suppliers format invoices differently, making traditional custom scrapers fragile and hard to maintain.
Invoices arrive as a mix of clean digital PDFs and scanned images, requiring intelligent routing between text extraction and vision OCR.
Automated workflows need webhooks that trigger actions as soon as a document is processed without manual exporting.

Document types this workflow handles

Digital GST invoice PDFs: e-invoices and billing software exports
Scanned invoices & photos: visual character recognition for images
Bulk billing exports: batch processing of text-layer invoices
Credit & debit notes: structured adjustment records

What comes out as JSON keys

Structured JSON. Keep these fields attached to their source. Illustrative workflow.

The product export provides reviewed columns, rows, and document metadata. Map these invoice values into the stable property names your application owns:

GSTIN (supplier): 29AABCC3456C1ZD
Invoice number: KT-5512
Invoice date: 2026-08-07
Vendor / customer: Ganesh Industries
Place of supply: Karnataka (29)
HSN / SAC code: 8467
Item & quantity: Rotary hammer x 6
Taxable value: ₹1,86,000.00
IGST (rate · amount): 18% · ₹33,480.00
Invoice total (incl. tax): ₹2,19,480.00

एक यथार्थवादी उदाहरण

Choose JSON for reviewed GST data. The export-menu library example uses a non-GST invoice.

Sample data. Not a real invoice or filing record.

The workspace provides structured JSON matching these reviewed table rows:

Invoice #	Date	Vendor	GSTIN	HSN/SAC	वस्तु	Taxable value	CGST	SGST	IGST	कुल
KT-5512	2026-08-07	Ganesh Industries	29AABCC3456C1ZD	8467	Rotary hammer	₹1,86,000	Not applicable	Not applicable	₹33,480	₹2,19,480
INV/2026-27/0188	2026-08-11	Mehta Steel Supplies	27AABCT1429P1ZM	7210	GI sheets	₹5,40,000	₹48,600	₹48,600	Not applicable	₹6,37,200

What to validate before export

Validation issue

Validate keys, numeric types, nulls, and tax totals against the source.

Limitation

The JSON is a product export, not a GST Portal return payload.

Reviewed by Dynamite Docs content team. Last verified 2026-09-30.

Reliable API integration and predictable data structures

Header fields, nested items. Check property names. Keep line items nested. Validate values and types. Illustrative workflow.

Keep document fields and line items separate. Supplier GSTIN, recipient GSTIN, invoice number, invoice date, place of supply, currency and invoice total describe the document. HSN or SAC, description, quantity, unit, rate, taxable value and tax components repeat by line. Flattening both levels without a stable invoice key can duplicate totals and corrupt later aggregation.

Treat identifiers as strings. GSTINs, invoice numbers, IRNs, HSN or SAC codes and purchase-order references may contain letters, slashes or leading zeros. Dates need a declared format, and decimal money values need an agreed representation. Your application should validate those types after extraction rather than relying on JavaScript coercion.

Define null and absence deliberately. A missing IGST value is different from a printed zero, and a field that is not applicable is different from one OCR could not read. Map these states into the application contract so reviewers and downstream rules can tell them apart.

Validate arithmetic before delivery. Compare line taxable values with the document subtotal, then account for discounts, freight, tax, cess and rounding. Keep the source amount when the calculation differs and record an exception. Do not silently replace a printed value with a computed one.

API and webhook consumers also need operational controls. Authenticate requests, verify the webhook signature, reject stale or duplicated deliveries, and store the extraction or file identifier used for idempotency. A completion event should start validation or review, not create an accounting voucher without checks.

Use confidence as a review signal, not an approval rule. GSTIN roles, tax treatment and totals deserve explicit checks even when OCR is confident. Route unresolved fields to a person and preserve their corrections before the application maps the JSON into its own schema.

Test the integration with a normal invoice, a credit or debit note, a multi-rate document and a scan. Version your mapper when required keys or nesting changes. Dynamite Docs supplies reviewed extraction data; your software remains responsible for schema validation, posting rules, duplicate control and GST compliance. Use the free invoice-to-JSON tool when you want to inspect one non-GST or prototype payload before building the API path.

Developer-friendly APIs with complete data control

The export uses a consistent product result shape for document metadata, columns and rows. Downstream code should map and validate that result against its own versioned schema.

A digital PDF may provide machine-readable text, while scans need visual OCR. Source quality can still change the proposed values, so both formats need review.

All data and webhook configurations remain under your control, with full support for CSV and Excel exports whenever needed.

सवालों के जवाब

Is there a REST API for uploading invoices and retrieving JSON?

Yes. Our REST API allows you to upload documents securely using API tokens and receive structured JSON responses with confidence scores for each field.

Do webhooks fire when invoice extraction is complete?

Yes. You can configure signed webhooks that post extraction completion events directly to your server endpoints, enabling fully automated workflows.

Are JSON key names consistent across all suppliers?

The product export uses a consistent result structure for fields, columns and rows. Your application should map those labels into its own schema and validate required keys, types, nulls and version changes.

Does the API support both digital PDFs and scanned images?

Yes. Digital PDFs parse directly from text, while scans and photos are routed through visual OCR, with both paths returning the same structured JSON format.