Receipt OCR built for accountants, not couriers.
Most OCR was designed for one receipt at a time. Accountants do not work that way. Pileform reads the whole pile, applies VAT regime defaults, groups by supplier, and hands back post-ready Excel workbooks with the original receipts embedded.
Pure OCR gives you text and stops. The accountant then has to manually decide whether “19%” printed under one line applies to that line or to the supplier as a whole. Whether the total is inclusive or exclusive. Whether the 5¢ gap is rounding or a typo. Whether 47 fuel receipts belong on one tab or 47 tabs. That post-OCR work is most of the job, and generic tools leave you doing it manually.
Pileform is OCR plus the accounting pipeline that comes after it.
From PDF to post-ready, no manual cleanup.
Every stage between “raw OCR text” and “workbook your client can file from,” built once so you do not have to rebuild it on every quarter-end.
OCR with layout understanding.
Multi-column receipts, mixed orientations, faded thermal prints, photos taken at an angle. The OCR pass extracts text, but the layout model understands which text belongs to which line.
VAT regime aware.
Cyprus 19/9/5/0, UK 20/5/0, EU multi-rate, GST regimes. The rate column is per-line. When the rate is not printed, Pileform infers from supplier category and yellow-flags the row for review.
Inclusive vs. exclusive detection.
“VAT included” in English, “συμπεριλαμβανομένου ΦΠΑ” in Greek, “TVA incluse” in French, 8 more. Every total is tagged with whether VAT is already inside it, with the source phrase preserved for audit.
Supplier grouping.
Forty-seven fuel receipts consolidate into one Petrolina tab with the supplier-level VAT control total. That tab is the unit your accounting software expects, so posting becomes one drag, not 47.
Cash rounding handled.
The Adjustment column captures any rounding gap (Cyprus 5¢, Dutch 5¢, Hungarian 5 forint, etc). The workbook always reconciles. Nothing gets silently fudged.
Yellow-flag for uncertainty.
When Pileform is not sure (faded thermal print, handwritten ink, ambiguous rate), the row renders with a citron flag and the source word preserved in a cell comment. You eyeball 4 rows, not 200.
Original receipt embedded.
Each row links to the original page image, embedded inside the workbook itself. When the auditor asks where a €342 line came from, you click the row and see the actual receipt. The workbook is self-contained.
.xlsx out, or post to your books.
You always get the .xlsx. Connect Xero or QuickBooks and the entries you confirm post straight through: Bills in Xero, journal entries in QuickBooks. If you stop using us tomorrow, you keep every workbook. No lock-in.
See a sample workbook before you sign up.
The features page has a real (anonymised) workbook with the rate column, the Adjustment cell, and the embedded receipts. Look first; commit second.
Receipt OCR questions.
OCR is the floor, not the product. Pure OCR gives you text. Pileform gives you structured accounting data: supplier, VAT number, date, line items, rate per line, currency, payment method. Then it groups by supplier and applies regime defaults. The OCR is just step one of a longer pipeline.
Pileform extracts handwritten receipts with lower confidence. The row is yellow-flagged for review before posting. Validating a flagged row takes seconds; typing one from scratch takes minutes. The flag is the contract: we tell you exactly which rows to eyeball.
Generic OCR returns text. Acrobat returns text with some table-detection. Neither knows VAT regimes, inclusive vs. exclusive phrasing, supplier grouping, or cash rounding. The output from those tools is the start of your work, not the end of it. Pileform’s output is post-ready: per-supplier tabs with VAT control totals and embedded receipt images.
Yes. Connect Xero or QuickBooks and the entries you confirm post straight through: supplier invoices as Bills in Xero, journal entries in QuickBooks. Your chart of accounts syncs in, and bank-statement transactions post as double-entry too. You still get the .xlsx workbook, and you are never locked in. Sage isn't supported. Tell us if you need it.
Accuracy depends on the source. Clean printed receipts extract reliably. Smudged thermal prints and handwritten slips are harder. Rather than quote a single number, Pileform returns rows in two states: high-confidence rows ready to post, and yellow-flagged rows for human review. Your time goes to the second category, not to every line.
Stop typing receipts by hand.
Sign up free, no card. Drop your messiest receipt PDF and see what comes back.