Blog
Notes on document extraction
Practical writing on schemas, accuracy, infrastructure, and the unglamorous work that makes document extraction reliable.
-
How to stop hand-keying documents in your accounting workflow
Most finance teams still retype invoices, receipts, and bank statements into their accounting system by hand — line by line, vendor by vendor. Here's how to take the data-entry step out of accounts payable, expense reconciliation, and month-end close without changing the tools you already use.
The Ztract team 8 min read -
OCR vs document extraction — why characters aren't data
OCR turns a scanned page into text. Document extraction turns it into fields you can use — invoice_number, total, line_items — each tied back to where it came from. If you've ever run OCR and still had to retype everything into a spreadsheet, this is the difference that matters, and how to tell which one your workflow actually needs.
The Ztract team 8 min read -
How to extract invoice data to Excel — whatever the layout
Every vendor sends invoices in a different layout, and that's exactly what makes them painful to get into a spreadsheet. Here's how to pull invoice number, dates, totals, and line items into clean Excel, CSV, or JSON — across any layout, without building a template per supplier.
The Ztract team 7 min read -
How to write a good extraction schema in plain language
The schema is just a description of what you want pulled out of a document. You don't need to learn a syntax to write one — but a few habits make the difference between output you trust and output you have to babysit. Here's how to describe what you want so the engine gets it right the first time.
The Ztract team 7 min read -
How to turn bank statements into clean Excel — without typing them by hand
Bank statements are some of the messiest documents to digitize: every bank lays them out differently, tables run across pages, and a misread minus sign throws off the whole reconciliation. Here's how to extract them into clean Excel, CSV, or JSON.
The Ztract team 7 min read -
Welcome to Ztract — and what we're building
We started Ztract because we kept watching people retype data out of documents that a computer could have read for them. This is what we're shipping, and what we're not.
The Ztract team 3 min read
