Invoice data extractiondown to the line item.
Extract data from invoices in any layout and language: header fields, totals that are recalculated until they add up, and line items from complex tables. No templates, no per-supplier setup.
Finance teams post invoices the day they arrive, instead of typing them in line by line.
What MiruIQ extracts from an invoice.
The typical invoice schema. Every field is yours to rename, drop or extend.
| Field | What it captures |
|---|---|
| Invoice number | The document's unique identifier, wherever the layout puts it. |
| Invoice date | Issue date, normalized to a standard date format. |
| Due date | Payment deadline, explicit or derived from payment terms. |
| Supplier | Issuing company name and address. |
| Supplier VAT / tax ID | VAT number or tax identifier, validated for format. |
| Buyer | Billed party name and address. |
| Currency | Invoice currency, normalized to ISO codes. |
| Net total | Total before tax, cross-checked against the line items. |
| Tax rate and amount | Per-rate tax breakdown; computed amounts are verified. |
| Gross total | Final amount, recalculated from net plus tax. |
| Payment terms and IBAN | Terms, discounts and the payable-to bank account. |
| Line items | Description, quantity, unit price and amount per row, from simple lists to complex nested tables. |
Not a pre-trained model, a schema you control.
These fields are a starting point, not a catalog: extraction is generic and schema-driven, so a field only your industry uses is one schema line away.
That is also why unusual invoices do not break it. Totals are validated by recalculation, and tables are read the way a human reads them, including notes above a table that change how the numbers must be interpreted.
Invoice extraction questions
What teams evaluating invoice data extraction ask most.
You define the fields you need as a schema, and extraction finds them by meaning rather than position: no templates, no per-supplier training. Extracted totals are recalculated, tax rates checked and IDs validated; invoices that fail a check park for human review. Structured results deliver to your systems by webhook, database or S3.
Yes. Line items extract row by row even from dense, multi-page or nested tables, and interpretive notes are respected: if a note above the table says amounts are in thousands or that zeros are cut off on purpose, the extraction reads the numbers accordingly instead of taking them literally.
No. Extraction is not pre-trained on document types, so an invoice from a new country, supplier or format processes the same way as a familiar one. The schema describes what you need; the layout is the document's problem, not yours.
See it on your own documents.
The 14-day free trial covers 200 documents: enough to define your schemas, run real files and measure the results before anyone calls you.