Any document in
Digital PDFs and scans or photos alike. Bank statements, invoices, receipts — anything laid out in columns. Open a file, or click “See an example” to try it instantly.
Fulgoria
Extract data from any document.
Fulgoria turns documents that are meant to be read into data you can use. You mark the columns of a document — a bank statement, an invoice, a receipt — once, and from then on Fulgoria recognizes every document of that kind and hands you clean, structured data as CSV.
It starts with bank statements, where a built-in balance rule
(opening + credits − debits == closing) gives a degree of certainty no extractor relying
on guesswork can match, and generalizes to any tabular document. Everything runs inside
your browser: the document never leaves your machine, and there is not a single external
request. Fulgoria is a standalone satellite of the Escriba ecosystem.
What it does
Section titled “What it does”Local OCR
Scans and images are read with Tesseract.js, running in the browser. No upload, no third-party OCR service.
Mark columns once
Automatic column detection does the first pass; you fine-tune by dragging. Assign a role to each column — date, description, debit, credit, amount with sign, balance, text.
The balance rule
For bank statements, Fulgoria confirms in green that it read correctly when
opening + credits − debits == closing. No votes, no heuristics you have to trust — just arithmetic, fully local.
Reusable templates
Mark a bank once; next time it is recognized automatically. The template stores where each datum lives, never what it says.
Spreadsheet-style editor
Fix any cell before exporting — find & replace, add or delete rows, paste from Excel, undo/redo, fill-down, assisted categorization and quality flags, with the balance recomputing live.
Multi-page, multi-account
Long statements and mixed-currency PDFs are handled — pesos and dollars in one file become one card per account.
Export to CSV + template
Export the movements as CSV, the geometry as a reusable .ext.json / .ext.yaml
template, or hand the result straight to Escriba.
Private by design
The PDF is opened, read and processed inside your machine. Fonts and libraries are vendored — privacy is the architecture, not a promise.
Part of the Escriba Suite
Section titled “Part of the Escriba Suite”Once you have extracted a document, “Send to Escriba” hands the clean result to your local Escriba to anonymize, convert (JSON / YAML / TOON), chunk for RAG, or turn into audio — and the document still never leaves your machine. Fulgoria stands alongside Fisherboy as a standalone satellite of the Escriba ecosystem.