GSTBot: AI‑Powered GSTR‑2B Matching for GST Reconciliation

GSTR‑2B Matching

The Problem

When a business files GSTR‑2B, it receives a consolidated list of all invoices that suppliers have reported. For small and medium‑sized enterprises (SMEs) in India, reconciling this list against the company’s own purchase register is a manual, time‑consuming process that often goes unnoticed until a tax audit surfaces discrepancies. Missing invoices, duplicate entries, or mismatched amounts can lead to incomplete Input Tax Credit (ITC) claims, higher compliance costs, and possible penalties.

SMB owners, accountants, and CA practices typically rely on cloud services such as ClearTax Pro or Taxilla to perform this matching. However, those solutions are priced beyond the reach of many small firms. GSTBot was built to fill that gap, delivering a robust GST reconciliation workflow at a fraction of the cost.

How GSTBot’s AI‑Driven Matching Works

1. Invoice Ingestion – GSTBot accepts invoices as images, PDFs, or Excel files. A lightweight OCR pipeline powered by Tesseract extracts structured data: vendor GSTIN, invoice number, invoice date, taxable amount, tax rate, and HSN code.

2. Data Normalisation – The extracted fields are normalised into a common schema. During normalisation, GSTBot verifies that the invoice number follows the standard 15‑character format and that the GSTIN matches the supplier’s registered number.

3. GSTR‑2B Retrieval – For each registered GSTIN, GSTBot pulls the latest GSTR‑2B filing from the GST portal via the చోట. The data is cached in Redis to avoid repeated API calls, improving performance for repeated reconciliation runs.

4. Reconciliation Engine – The core engine compares each invoice record against the GSTR‑2B dataset. The algorithm considers:

   • Invoice number and date – a primary key for matching.
   • Taxable value and tax amount – ensuring values match within a tolerance of ±₹5.
   • HSN code – to flag mismatches that could affect ITC eligibility.

When a record does not find a corresponding GSTR‑2B entry, GSTBot marks it as unmatched and logs the reason (e.g., missing invoice, duplicate, amount mismatch).

1. ITC Eligibility & Leakage Prevention – For matched invoices, GSTBot calculates the eligible ITC. It automatically flags invoices that violate Rule 37, 42, or 43 (e.g., reverse charge, taxable supplies from outside India) and prevents ITC claim for those entries. The system also highlights potential ITC leakage by comparing the total eligible ITC against the amount reported in GSTR‑3B.

2. Supplier Reliability Scoring – Each supplier receives a reliability score based on their filing consistency and compliance with tax rates. GSTBot surfaces suppliers with a low score, prompting the user to investigate or consider alternative vendors.

Using the Feature

1. Upload Invoices – From the dashboard, the user selects the “Upload Invoices” button and drags files into the drop zone. GSTBot immediately starts OCR and data extraction.

2. Review Extraction – A preview table shows each extracted field. The user can manually correct any mis‑parsed values before proceeding.

3. Run Reconciliation – Clicking “Reconcile” triggers the AI engine. GSTBot displays a progress bar that updates as each invoice is compared against GSTR‑2B.

4. View Results – The results page lists three sections:

   • Matched – invoices that align with GSTR‑2B, with a green checkmark.
   • Unmatched – invoicesISC flagged with a red exclamation point and a tooltip explaining the mismatch reason.
   • ITC Eligibility – a summary table showing the total eligible ITC, the amount that will be reported in GSTR‑3B, and a warning if there is a discrepancy.

5. Export – The user can export the reconciliation report as JSON or CSV. The JSON format is GST‑portal compatible, enabling a one‑click pre‑fill into GSTR‑1 or GSTR‑3B.

Impact for Indian SMBs

By automating the most error‑prone part of GST compliance, GSTBot reduces manual effort from days to minutes. The real‑time flagging of missing or duplicate invoices prevents ITC leakage, while the supplier reliability score encourages better vendor relationships. Because the entire workflow runs on the open‑source stack of FastAPI, React, PostgreSQL, and Redis, the cost of ownership remains low, making high‑level AI GST compliance accessible to every small business.

In short, GSTBot’s GSTR‑2B matching feature turns a cumbersome reconciliation task into a fast, reliable, and transparent process that safeguards ITC claims and keeps SMBs compliant without the premium price tag of larger platforms.