GSTR-2B reconciliation document processing

Input tax credit reconciliation starts with two lists: the GSTR-2B statement, the auto-drafted input tax credit statement generated on the GST portal, and your own purchase invoices and receipts. Matching them means comparing supplier GSTINs, invoice numbers and dates, taxable values, and the CGST/SGST/IGST split across both, and most of that comparison still starts by retyping the purchase side out of PDFs. Dynamite Docs processes purchase invoices and receipts into structured rows — supplier GSTIN, invoice no/date, HSN/SAC, taxable value, tax split — so reconciliation against your GSTR-2B data starts from clean rows instead of a wall of PDFs.

Try this workflow free or view pricing.

Input credit reconciliation starts with retyping

Every purchase invoice is a small reconciliation task: supplier GSTIN, invoice number and date, taxable value, and the CGST/SGST/IGST split all have to be checked against the GSTR-2B statement line by line.
Purchase documents arrive in every shape, billing-software PDFs, scans, photos, and receipts, and the fields repeat on every one with a different layout.
The mismatches are where the work hides: an invoice in your records but not in the statement, a GSTIN mismatch, a tax amount that doesn’t tie, and each one has to be traced back to the original document.
The purchase side of the match is almost always manual, because the statement’s data is structured but the underlying invoices are still a stack of PDFs.

How Dynamite Docs handles GSTR-2B reconciliation documents

01 — Drop the purchase batch in

Upload a folder of purchase invoices and receipts, or drag in the PDFs and scans vendors actually send. Clean digital PDFs are parsed deterministically from their text layer: no AI call, identical results every time.

02 — Schema inferred, not built

Supplier GSTIN, invoice number and date, HSN/SAC, taxable value, CGST, SGST, IGST, and totals are read from each layout on the spot. No template library to maintain for every vendor format.

03 — Review by confidence

Every field carries a confidence score. You check the low-confidence values, usually a GSTIN or a tax split that doesn’t tie to the taxable value, and each correction becomes a pattern the next batch reuses.

04 — Export and match

Convert rows to Excel, CSV, or JSON. With the purchase side structured, the match against your GSTR-2B data becomes a comparison of rows instead of a retyping exercise.

Document types this workflow handles

Purchase invoices — supplier bills for goods and services
Supplier receipts — delivery and payment receipts
Credit notes from suppliers — vendor adjustments and corrections
E-way bills — transport documents alongside the invoice

What gets extracted from a purchase document

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

GSTIN (supplier) — 27AABCS1429B1ZP
Supplier name — Shri Balaji Traders
Invoice number — PV/24-25/0088
Invoice date — 2026-08-02
HSN / SAC code — 4820
Item & quantity — A4 copier paper × 40
Taxable value — ₹24,000.00
CGST amount — ₹2,160.00
SGST / IGST amount — ₹2,160.00 / ₹33,480.00
Invoice total (incl. tax) — ₹28,320.00

A realistic example

A batch of purchase documents extracts to rows like these, each row one invoice, receipt, or adjustment ready to compare against your GSTR-2B data:

Invoice #	Date	Supplier	Supplier GSTIN	HSN/SAC	Taxable value	CGST	SGST	IGST	Total
PV/24-25/0088	2026-08-02	Shri Balaji Traders	27AABCS1429B1ZP	4820	₹24,000	₹2,160	₹2,160	—	₹28,320
PV/24-25/0091	2026-08-05	Kaveri Engineering Co.	33AABCC3456C1ZD	8467	₹1,86,000	—	—	₹33,480	₹2,19,480
CR/24-25/0017	2026-08-07	Shri Balaji Traders	27AABCS1429B1ZP	4820	₹4,000	₹360	₹360	—	₹4,720

The two lists behind input credit reconciliation

GSTR-2B is an auto-drafted statement of input tax credit available, generated from the outward supplies reported by your suppliers. Reconciliation means comparing that statement’s line items against your own purchase records so you can see what matches, what’s missing, and what needs follow-up with a vendor. The comparison is a bookkeeping task, and it starts with the purchase documents. Dynamite Docs turns those documents into rows the moment they are uploaded, so the match runs on data instead of retyping.

The procedural part stays outside the tool. Whether a credit is claimable, how a mismatch should be resolved, and when to follow up with a supplier are decisions for your CA and your team, and the rules change — consult the GST portal and your tax advisor for current requirements. What extraction removes is the keystrokes: the supplier GSTIN, invoice number and date, taxable value, and tax split arrive confidence-scored and ready to compare.

The tax split is where the reconciliation actually bites. CGST and SGST apply to intra-state supplies and IGST to inter-state supplies, and a row that should have carried IGST but was keyed as state tax will surface as a mismatch. Because the split is extracted as separate fields and checked against the total, the bookkeeper’s review shrinks to the handful of rows that genuinely don’t tie.

Volume is the argument for batching. A practice reconciling purchase invoices for several client firms processes hundreds of documents a month, and every retyped number is a small risk of a mismatch nobody wants to chase. Batch processing plus per-layout pattern memory means the supplier that sends the most invoices extracts cleanly after the first correction session.

Why bring-your-own-AI matters for reconciliation data

Reconciliation is a comparison, and the comparison is only as good as the purchase side it starts from. Supplier GSTIN, invoice number and date, taxable value, and the tax split have to be exact on every bill, and the model that reads the scanned tenth of that volume is your choice, paid at your provider’s rate with zero markup.

Clean purchase invoices, including most billing-software exports, parse deterministically for free, identical results every run. Vendor scans and photos are read by the vision model you pick, so the cost of the month’s reconciliation input stays under your control.

With the purchase side structured, the match against your GSTR-2B data runs on rows you can filter and sort. The CGST/SGST versus IGST split surfaces mismatches early, corrections per vendor compound, and the exported rows stay yours for the ledger.

Questions, answered

Does Dynamite Docs pull GSTR-2B from the GST portal?

No. It processes your purchase invoices and receipts into structured rows. The GSTR-2B statement data comes from your own access to the portal; the tool gives you the structured purchase side to match against it.

Does it decide whether input tax credit is available?

No. Credit eligibility and claim decisions rest with your CA and your team. Extraction hands you the supplier GSTIN, invoice number and date, taxable value, and tax split; the judgment stays with the people responsible for the books.

Can I export purchase rows to Excel, CSV, or JSON for the match?

Yes. Export to Excel, CSV, or JSON, or push rows to Google Sheets, so the comparison with your GSTR-2B data runs on rows you can filter, sort, and reconcile.

How does the CGST/SGST/IGST split come out?

Taxable value, CGST, SGST, IGST, and the invoice total come out as separate confidence-scored fields, so the split can be checked against the total before anything reaches your reconciliation.