Which file types are supported?

PDF, DOCX, PPTX, XLSX, TXT, Markdown, EPUB and HTML, uploaded or fetched from a link.

Does it handle scanned PDFs?

Yes, on Pro and API: pages without a text layer are OCR'd at 3 units a page. The free plan extracts the text layer only.

No. Files are processed in memory and discarded; the extracted text is kept in your history so you can come back to it.

How big can a file be?

25 MB on the free plan, 200 MB on Pro and API.

Free to start

Document Extractor

Turn any document into clean text, page-anchored Markdown, tables and outlines you can search, cite and feed to an AI.

Extract a document See pricing and plans

Published 24 September 2026 · Reviewed 24 September 2026

Who it is for

Researchers and students
Consultants and analysts
Developers building RAG pipelines
Anyone with a scanned PDF

The problem

Text copied out of a PDF loses headings, breaks mid-word, scrambles tables and carries no page numbers, so an AI summary cannot say where a fact came from. Scans have no text layer at all.

How GistCite solves it

Upload the file or paste a link. Plain text comes back immediately from pypdf; a layout-aware Markdown conversion by Docling follows in the background with a page anchor before every page, a headings outline and tables you can export as CSV or XLSX. Scanned pages go through OCR on paid plans.

Capabilities

Immediate text, anchored Markdown

Plain text first; Markdown with page anchors and headings a moment later.

Benefit: Summaries and answers cite the page they came from.

Extract a document to Markdown

Tables and outlines

Tables are detected and exportable as CSV or XLSX; headings form a clickable outline.

Benefit: Get the numbers out of a report without retyping them.

How table and outline extraction works

OCR for scans and photos

Pages without a text layer are recognised at 3 units a page (Pro).

Benefit: Receipts, old reports and photographed pages become searchable.

OCR limits and cost

Batches and translation

Drop up to 25 files for one ZIP, or translate a document into 30 languages with structure kept (Pro).

Benefit: Process a folder, not a file.

See batch and translation on the Pro plan

Nine export formats

PDF, DOCX, PPTX, XLSX, image, TXT, Markdown, EPUB and HTML, plus RAG-ready chunked JSON with offsets.

Benefit: The output lands in whatever tool comes next.

Every export format explained

Benefits

Files are processed in memory and never stored; only the extracted text is kept for your history.
Free plan accepts files up to 25 MB; Pro and API up to 200 MB.
The same extraction is one API call, so a pipeline can run it on every new file.

How it works

Step 1
Upload or link
PDF, DOCX, PPTX, XLSX, TXT, MD, EPUB or HTML, from your computer or a URL.
Step 2
Extract
Text immediately; Markdown, outline and tables in the background; OCR when a page has no text.
Step 3
Export or continue
Save in nine formats, or summarize, ask, translate and create from it.

Integrations and formats

REST API
CSV / XLSX table export
Chunked JSON for vector stores
Markdown Editor
AI Summary, Ask, Create and Translate

Security and data handling

Read the full picture on the security and compliance page.

Uploads are processed in memory and never written to disk or object storage; only the extracted text is kept.
Every extraction endpoint is rate-limited per IP; signed-in use is metered against the plan's monthly quota.
TLS in transit; passwords bcrypt-hashed; API keys stored only as SHA-256 digests.

Evidence

Extraction runs on pypdf for immediate text and Docling for layout-aware Markdown, both open-source. Check this
Every page gets a `` anchor so citations point to a real page. Check this

Worked examples

Independent consultant

From a 120-page annual report to a cited two-page brief

A public annual report PDF becomes page-anchored Markdown, a set of extracted tables, answers with page citations, and an executive brief a client can verify.

Read the independent consultant walkthrough

Graduate student

From a twelve-lecture playlist and two textbooks to a searchable study corpus

A public lecture playlist and two public-domain textbooks become one set of transcripts, summaries and chunked JSON, ready for revision or a personal AI index.

Read the graduate student walkthrough

Pricing

Plans: Free, Pro and API
Cost per action: 1 unit per document; OCR 3 units per page; translation 2 units per page; batch items 1 unit each.
How it is priced: File size limits are 25 MB on Free and 200 MB on paid plans. The cost of an action is shown before it runs.

Compare the Free, Pro and API plans