Bank Statement OCR API

4.8/5

(30+ reviews)

Bank statement OCR API that reads every transaction

Turn PDF statements, scans and phone photos into structured JSON in seconds. Account holder, period, balances and every transaction line, each with a confidence score. No template per bank, no rules to maintain.

No credit card required

SOC 2 Type II & GDPR

Bank statement OCR extracting account holder, balances and transactions to JSON

Trusted by finance and product teams at

Why bank statements are slow to review

Every bank formats its statements differently, and a single file can run to dozens of pages of transactions. Analysts end up retyping balances and scanning lines by hand to check income and spending.

Statement uploaded

Months of PDFs, scans or photos from the applicant.

Manual review

Balances and transactions read page after page.

Friction

Every bank lays out its columns differently.

Income check

Salary credits and recurring debits added up by hand.

Friction

One missed line skews the affordability check.

Cross-check

Totals compared with payslips and the application.

Friction

Mismatches send the file back to the applicant.

Decision

Finally approved, days after the documents arrived.

RESULT

[X] minutes to review each statement by hand, and [X]% of loan files delayed by a missing or unreadable page.

See how Mindee fixes it

Account holder

account_holder_names, account_holder_address

Account

bank_name, account_number, account_type, branch_code, currency

Period

statement_period_start_date, statement_period_end_date, statement_date

Balances

beginning_balance, ending_balance, total_credits, total_debits

Transactions

date, description, amount

Daily balances

date, balance_amount

response.json

{

"account_holder_names": ["EMMA CARTER"],

"bank_name": { "value": "Harbor Bank" }, "account_number": { "value": "****4821" }, "currency": { "value": "USD" },

"statement_period_start_date": { "value": "2026-09-01" }, "statement_period_end_date": { "value": "2026-09-30" },

"beginning_balance": { "value": 2140.55 }, "ending_balance": { "value": 2896.10 }, "total_credits": { "value": 3410.00 },

"list_of_transactions": [ { "date": "2026-09-25", "description": "SALARY ACME CORP", "amount": 3250.00 }, … 41 more ],

"daily_balances": [{ "date": "2026-09-01", "balance_amount": 2140.55 }, …],

}

Ready to post, not just to read.

Dates come back in ISO 8601, amounts and balances as numbers, addresses split into street, city and postal code. Every field carries a confidence score.

Built for real statements, not clean exports

Scanned pages with a stamp, phone photos of a printout, 20-page PDFs with tables that run across page breaks. The model reads statements the way customers send them.

Statements from five banks in different currencies read into one JSON schema

Any bank, one schema

A statement from a bank you have never seen comes back in the same JSON shape. No template to add per bank or country.

Transaction table running across a page break, returned as one list of 42 lines

Every transaction, as its own line

Transactions come back as an array with date, description and amount, even when the table runs across dozens of pages.

Opening balance plus credits minus debits matching the closing balance

Totals you can reconcile

Opening and closing balances, total credits and total debits come back next to the lines, so your code can check they add up.

Daily balance chart for one month with the lowest point highlighted

Daily balances included

A balance for every day of the period, ready to spot overdrafts or low points without recomputing anything.

PDF / JPEG / PNG / WEBP / TIFF / HEIC, Multi-page statements, Any bank / any currency, Scans & phone photos, Stamped & annotated pages

Bank statement OCR accuracy, checked line by line

Each value comes with a confidence score, so clean statements flow straight into your checks and doubtful lines go to an analyst. Test it on your own statements in the free trial.

95.6%

Field-level accuracy on header fields

Holder, account, period, balances

98.2%

Accuracy on transaction lines

Date, description, amount

0.7s

Median time per statement

Three-page PDF, synchronous call

A confidence score on every field

Set your own threshold. Fields above it flow straight to your systems; anything below lands in a review queue, with the exact spot on the bank statement highlighted. Définissez votre propre seuil. Les champs au-dessus partent directement dans vos systèmes ; les autres rejoignent une file de vérification, avec leur emplacement exact mis en évidence sur votre bank statement. Define tu propio umbral. Los campos que lo superan pasan directamente a tus sistemas; los que quedan por debajo van a una cola de revisión, con la ubicación exacta resaltada en bank statement.

Auto-approve the bank statements you can trust Aprueba automáticamente bank statements de confianza Validez automatiquement les bank statements fiables

Send only real exceptions to a human

Document AI features, better than OCR

Prepare, classify and extract. Mindee delivers high-accuracy extraction for every document format, so you can automate workflows with speed and high reliability.

Prototype in minutes, go live with the API

Start with a no-code workflow to prove the value on your own bank statements, then move the same model into your product. Commencez par un workflow no-code pour prouver sa valeur sur vos propres bank statements, puis intégrez le même modèle dans votre produit.

What teams build with bank statement OCR

The same API powers very different products. Here is where it usually starts.

Rated 4.8/5 by developers and lending teams

4.8

on G2

4.9

on Capterra

quote

Mindee is a web based tool that help us in scanning and reading different type of documents like identity cards, invoices, proposal plans etc and extract all the information with its AI and then it provides all the information and data associated with these documents a structured way.

on G2

quote

Mindee is a software that helps us to convert all of our physical business data like bills, invoices, warranty cards, calendar, recipts received to us into a digital documents that can be stored in our drive and can be uploaded in different type of Excel sheets so that all the updates can be maintained and a proper analytics of transactions can be kept by the financial team

on G2

quote

Excellent. In addition to their great product, the sales team has always been proactive on how they could help us leverage the maximum results from their product. It was like having an additional product manager on our side

on Capterra

FAQ about Mindee's Bank Statement OCR API

Can't find what you're looking for? Our team answers within one business day.

About bank statement OCR À propos de l’OCR pour les bank statement

How do I convert a bank statement PDF to Excel or CSV?

Send the PDF to the Mindee API and it returns every transaction as structured data, with the date, description, amount (debit or credit) and running balance for each line. It also returns the statement details: account holder, bank, account number, statement period, and opening and closing balances. Because the output is JSON, you can export it to Excel, CSV or Google Sheets in a few lines of code, or with no code at all through Zapier, Make or n8n. Unlike copy-pasting or generic PDF converters, rows stay intact, with no merged lines or shifted columns.

Does it work with multi-page statements from any bank?

Yes. Mindee doesn’t rely on bank-specific templates, so it handles statements from banks worldwide, whatever their layout, language or currency. It reconstructs transaction tables that run across dozens of pages, ignores repeated headers and footers, and correctly handles descriptions that wrap onto several lines. It works with both native PDFs downloaded from online banking and scanned or photographed paper statements.

Can Mindee help detect altered or fake bank statements?

Mindee is not a forensic fraud detection tool, but structured extraction makes several checks easy to automate. You can verify that the opening balance plus all transactions equals the closing balance, that each running balance is consistent from line to line, and that the account holder’s name matches the applicant. When a statement has been edited, these calculations often stop matching. Combined with confidence scores, these checks let lenders and fintechs automatically approve consistent statements and send suspicious ones to manual review.

Why not use ChatGPT to extract transactions from a bank statement?

General-purpose LLMs struggle with long transaction tables. On a statement with hundreds of lines, they can skip rows, merge two transactions or produce an amount that doesn’t exist, and nothing in the output warns you. For financial data, one missing line means the balances no longer reconcile. Mindee returns every transaction in a consistent JSON structure, with a confidence score and its position on the page. Bank statements also contain sensitive financial data: Mindee is SOC 2 Type II certified and GDPR compliant, which a general chatbot doesn’t guarantee for your documents. More details on Mindee vs LLMs.

About Mindee

Is my data secured and protected during the conversion process?

Yes, your data is highly secured and protected during the entire conversion process. As an enterprise-grade OCR solution, Mindee takes data privacy very seriously. Here are the key security measures in place:

  • Certified compliance: Mindee's infrastructure is SOC 2 Type II certified and fully GDPR compliant, ensuring strict industry standards for data protection.
  • Encryption: All documents and extracted data are fully encrypted both in transit (during the API call) and at rest.
  • Data localization: You have control over where your data is processed, with the option to host and process your data on either EU or US servers.
  • Strict privacy: Mindee respects data confidentiality. Your processed documents are not shared with third parties, and your private data is not used to train global AI models.

In short, your sensitive documents are processed in a secure, isolated environment and remain completely under your control.

Is Mindee's OCR API free to use?

We provide a 14 days free trial so you can fully test any of our OCR models — no credit card required. After this, we offer different pricing tiers depending of the volume of pages processed and the features you might need. See the pricing page for more information.

What file formats are supported for JSON conversion?

The supported document formats are:

  • PDF (application/pdf) – While technically a document format rather than an image, it is natively supported for all extraction models.
  • JPEG / JPG (image/jpeg)
  • PNG (image/png) – Must be non-animated.
  • WebP (image/webp)
  • TIFF / TIF (image/tiff) – Both single-page and multi-page TIFFs are supported and processed similarly to PDFs.
  • HEIC (image/heic) – Apple's High-Efficiency Image Container format.

‍

📌 Important technical limitations to keep in mind:

To ensure the OCR pipeline runs smoothly and successfully returns your JSON, make sure your files adhere to the following constraints:

  • Maximum file size: 100 MB per file.
  • Maximum page count: Up to 200 pages per document.
  • File state: The files cannot be encrypted, and PDFs must not be password-protected.

Ready to stop retyping bank statements?

Plug the bank statement OCR API into your lending or onboarding flow in minutes, with a team of document processing engineers behind it.

14-day free trial

No credit card required