Skip to content
Document intelligence · Built for real-world pages

Documents in.
Usable data
out.

Extract text, tables and mathematical formulas from PDFs and scans. See the result beside the original page — and check it before you use it.

See an example Open Foliandra

From scanned pages to structured, reviewable output.

research_report.pdfRecognition complete
Original document Extracted contentPage 03 / 12
Original PDF
Laboratory note · Sample

Material analysis

Methodology & observations — Section 2

SampleValueUnit
A–0112.6mg/L
A–0214.8mg/L
A–0313.1mg/L
E=mc2
Extracted Reading view
Table · illustrative output

Material analysis

| Sample | Value | Unit |
|--------|-------|------|
| A–01   | 12.6  | mg/L |
| A–02   | 14.8  | mg/L |
| A–03   | 13.1  | mg/L |
Formula · LaTeXE = mc^2

TXT · HTML · XLSX

Original & output, side by side
Illustrative document and output. Not an accuracy test.
Works withPDF · PNG · JPEG · TIFF · WEBPExport optionsTXT · HTML · XLSX · CSV · JSON/ZIP

Available formats depend on the recognition engine.

Simple process

From document to data,
in three clear steps.

Let the recognition engine read the page. Keep the final judgment in your hands.

01 — Upload

Bring your pages

Upload a PDF or a single-page image. Select the pages and choose a recognition engine.

02 — Verify

Compare with original

Inspect recognized text, table cells and formulas beside the source page. AI can make mistakes; you check the result.

03 — Save

Take your data

Choose from the save options for your engine: text, offline HTML, table files or source responses.

Beyond plain OCR

Complex pages deserve more than copied text.

A formula has symbols to check. A table has cells to compare. Inspect the result while keeping the original page in view.

  • Recognized tables you can inspect cell by cell
  • LaTeX source and visually rendered mathematics
  • Source responses kept separately from reading views
Sample extracted output

From table to reusable rows

Sample     Value       Unit
A–01       12.6        mg/L
A–02       14.8        mg/L
A–03       13.1        mg/L

AI OCR 1 can save recognized tables as XLSX or CSV. Table structure must be supported by the exporter.

From printed mathematics to LaTeX

E=mc2
E = mc^2

Compare rendered mathematics with the original and inspect the LaTeX source. Save options vary by engine.

From document pages to readable text

# Material analysis

Methodology & observations

The report describes three samples
and their measured values.

TXT and HTML are available for both engines. AI OCR 2 whole-run text exports exclude formulas and images.

Illustrative samples, not measured accuracy results.

Choose your approach

Two ways to read a page.

Choose preset-guided recognition or a fixed layout-aware process.

AI OCR 1

Prompt-guided reading

Choose a preset for the full page, main document content, text only, a table page or a formula image.

  • Flowing text, tables and formulas in reading order
  • Each page’s model response is retained
  • TXT and offline HTML; table exports in XLSX and CSV
AI OCR 2

Layout-aware reading

A fixed process identifies page regions — text, titles, tables, formulas and images — with their positions.

  • Compare the result in the page’s own layout
  • Save the current page as offline HTML
  • AI OCR 2 Cloud with your API token: PDF, PNG and JPEG

Save a reading view or source data using the options available for your result. AI OCR 2 whole-run TXT/HTML is project-derived text: formulas and images are excluded, and table text is optional.

Where it helps

Built for the difficult pages.

Useful when ordinary copy and paste is not enough.

01 / Structured data

Dense tables

Inspect recognized rows and columns, then use the table export options available for your engine.

02 / Scientific content

Formulas and research

Read mathematical notation as rendered formulas and LaTeX beside the printed original.

03 / Mixed layout

Scanned reports

Compare pages that combine headings, text, figures, tables and varied reading order.

Document handling

Know where your files go.

Choose your processing method deliberately. See the Privacy notice for document handling details.

A cloud provider receives the document only after you choose and confirm that run.

You control deletion of your uploaded documents.

Automatic cleanup after 10 days without opening; active runs and open support cases postpone it.

About

Who runs Foliandra.

Foliandra is an independent document-recognition service run by its developer. It is free to use during early access. Questions, problems or claims: sign in and open a support case in Account.

Foliandra never asks for your passwords to other services.

Nothing to install: Foliandra works in your browser.

Start with one document

Make your documents
work for you.

Explore Foliandra’s recognition and comparison workflow.

Open Foliandra