Extract Business Registration Numbers and VAT from Korean Tax Invoices and Cash Receipts Without Retyping Each One
Hometax (홈택스) covers electronic tax invoices, but the cash receipts, transaction statements, and paper invoices arriving as PDFs, scans, and photos still get typed by hand. Name the columns you need once, and this reads all of them in one batch.
5-10s per page · Up to 99% accuracy on printed Korean text
Every Korean Tax Document Type, One Set of Columns
A Korean tax invoice (세금계산서) and a cash receipt (현금영수증) share no visual structure, yet they share the same core fields. You name the fields you need once, and the AI finds each value wherever it sits on the page.
These are the column names you type, and the AI finds matching values regardless of format.
Korean Tax Documents Are a Family of Unrelated Formats
A printed tax invoice, a thermal cash receipt, and a government business registration certificate look nothing alike, yet all three carry the same business registration number (사업자등록번호). The AI finds that number wherever it appears, so format differences stop mattering.
Why Korean Tax Documents Break Traditional Extraction
Hometax (홈택스) covers only system-issued e-invoices. The National Tax Service portal handles electronic tax invoices transmitted through its own system, but the paper invoices, PDFs, photos, and scans from suppliers below the e-invoicing threshold, plus transaction statements that no government system receives, all stay manual.
The same field appears in unrelated layouts. The business registration number is a 10-digit identifier printed on tax invoices, cash receipts, and registration certificates, three documents with zero shared structure. Template-based tools must be reconfigured for each one; one format change breaks the match.
The field you actually need isn't printed anywhere. For year-end settlement (연말정산), the bottleneck isn't transcribing dates and amounts, it's the expense category, which no document prints. A year of collected cash receipts has to be sorted and categorized before the February deadline, and that sorting is where the hours go.
How the AI Reads Any Korean Document
Name the columns once, and the AI finds each value wherever it sits. Type "Business Reg. No., Supply Value, VAT" once. The model identifies the 사업자등록번호 from its 3-2-5 digit pattern and label relationship whether it appears on a government certificate, an ERP printout, or a scanned invoice, with no per-format template.
One batch for all document types. Upload tax invoices, cash receipts, transaction statements, and purchase orders into the same batch. Define columns like "Business Reg. No., Supply Value, VAT, Item, Qty, Total." The AI fills each document independently, and columns that don't apply stay empty. The export is one consolidated spreadsheet.
Let the AI fill in what no document prints. Define "Expense Category (options: 식대/교통비/사무용품비/임차료/통신비/소모품비/기타)" and the AI reads each receipt's content, determines the category, and fills it in. Extraction and categorization happen in the same pass, ready for your tax accountant (세무사).
From a Month of Mixed Korean Documents to a Filing-Ready Spreadsheet
Upload the Whole Mixed Stack
If you're reconciling a month of Korean supplier documents, you're holding tax invoice PDFs (세금계산서), thermal cash receipts (현금영수증), a transaction statement (거래명세서), and maybe a photographed paper invoice from a small vendor. Drag them all into one batch, PDF, JPG, PNG, or WebP, with no sorting by type required.
Name the Fields You Need Once
Type Business Reg. No., Date, Vendor, Supply Value, VAT, Total, Expense Category. One set of columns covers tax invoices, cash receipts, and transaction statements. The AI reads supply value (공급가액) and VAT (부가세액) figures from the invoice, the approval number and amount from the thermal receipt, and the item lines from the transaction statement, each routed to its named column.
Download One Filing-Ready Spreadsheet
Processing takes 5-10 seconds per page. The output is one XLSX or CSV: each row is one document, each column is a field you named. Invoice rows carry Supply Value and VAT, cash receipt rows carry Amount and Approval No., transaction statement rows carry Item/Qty/Total, and every row already has its Expense Category filled in (~18x faster than typing each value by hand).
When It Works Best, and Where to Check the Output
Extraction pulls the data off the document reliably, but filing accuracy is still your responsibility.
When It Works Best
Clear scans, PDFs, and photos of printed Korean documents. Machine-rendered tax invoices and registration certificates reach up to 99% accuracy on standard layouts.
Mixed document types in one batch. Tax invoices, cash receipts, transaction statements, and purchase orders process with one set of columns, no per-type configuration and one consolidated spreadsheet.
Standardized fields with consistent Korean labels. Business registration number (사업자등록번호), supply value (공급가액), and VAT (부가세액) use the same label across document types, which makes matching reliable.
When to Be Cautious
Heavily compressed or forwarded document photos. Images shared through messaging apps lose resolution, and faded thermal paper reduces accuracy on approval numbers and small print.
Fully handwritten Korean documents. A handwritten amount on a pre-printed tax invoice form works; irregular fully handwritten documents need manual review of extracted values.
This is extraction, not filing. Hometax (홈택스) already holds structured data for e-invoices, so no extraction is needed there. For everything else, spot-check a few rows before handing figures to your tax accountant (세무사).
Frequently Asked Questions
Can I extract data from a Korean tax invoice (세금계산서) and a cash receipt (현금영수증) in the same batch?
Yes. Define one set of column names (Business Reg. No., Supply Value, VAT, Total, Date) and the AI processes every document in the batch with the same definitions. A tax invoice fills Supply Value and VAT; a cash receipt fills Amount and its approval number. Columns that don't apply to a document stay empty for that row. The output is a single spreadsheet covering both document types.
Does it read Korean-language fields like business registration number (사업자등록번호) and supply value (공급가액)?
Yes. Korean-language support is built in, not layered on top of an English-only engine. The AI reads business registration numbers, supply values, VAT amounts, dates, and recipient information as they appear in Korean, then routes each value to the column you named. Printed and typewritten Korean documents reach up to 99% accuracy.
Can it handle a photographed or paper tax invoice, not just a PDF?
Yes. A clear photo of a printed tax invoice, a scanned copy, or a PDF all work, because the AI reads the visible document content from the image itself. For the best accuracy, photograph the document flat with even lighting and avoid cropping out the edges where business registration numbers and totals usually sit.
Will it extract the expense category even though it isn't printed on the document?
Yes. This is the one field no Korean tax system generates. Define a column like Expense Category (options: 식대/교통비/사무용품비/임차료/통신비/소모품비/기타) and the AI reads the document content, determines the correct category, and fills it in. Categorization happens in the same pass as extraction, so year-end settlement documents arrive classified, not just transcribed.
Does it output supply value (공급가액) and VAT (부가세액) as separate columns?
Yes. Name "Supply Value" and "VAT" as two separate columns and the AI splits them correctly. The 10% relationship between supply value, tax, and total is understood, not guessed. This matters for quarterly VAT filing, where supply and tax amounts are reported separately. The same understanding applies across all ten document types on this page.
Deep dives into Korean document extraction workflows: the step-by-step Korean tax invoice extraction guide, the workflow this page is built around · the complete guide to Korean cash receipt extraction for tax deductions · the Korea VAT deadline checklist for tax invoice data prep · how Korean document extraction pricing compares across government, ERP, and AI tools
If your task is filing the VAT return itself (the entire form as a full table rather than specific fields pulled out), VAT return form to Excel handles that different workflow.