DocTable
AI OCR online for PDF & images
Use DocTable as AI OCR / PDF OCR in the browser — no install. Multilingual, sign in once and scan.
Start — upload a documentWhat you can OCR
- Invoices & receipts — vendor, date, line items, tax and total as rows
- IDs and forms — label/value fields, not raw text
- Tables — digital or scanned, multi-column layouts included
- Letters and prose (Word export instead of Excel)
Also see PDF to Excel and invoice to Excel.
How AI OCR differs from classic OCR
Classic OCR engines recognize individual characters and hand back a wall of raw text — you still have to figure out which words were a table row and which were a header. DocTable uses a vision-language model instead: it reads the page the way a person would, identifying which pixels form a table, which are a label next to a value, and which are a stamp or signature. The output is structured rows and fields, not text you have to re-parse yourself.
What to expect from scanned PDFs and photos
Scanned pages, phone photos and screenshots go through the same vision OCR as digital PDFs — there's no separate "scan mode." Skewed or rotated photos and moderate glare are usually fine. Very low resolution, heavy motion blur, or cursive handwriting reduce accuracy; clear block handwriting on filled-in forms is often readable, but always verify handwritten fields manually before relying on them.
Supported languages
Persian, Arabic, Chinese, English, and mixed-language pages are all read by the same default OCR model — the model doesn't change based on which UI language you're browsing in, so switching the interface to Persian doesn't switch the OCR engine.
OCR scanner, OCR software, or OCR online — which do you need?
People search all three and mean different things. An OCR scanner usually means a phone app or a hardware scanner that captures the page in the first place. OCR software means something installed on a desktop, working on files already on that machine. OCR online means a browser tool with nothing to install and nothing tied to one computer.
DocTable is the third kind. There is no capture step and no mobile app: you bring an image or PDF you already have, and it runs in the browser on any machine you sign in from. If you still need to photograph paper, a scanner app handles the capture and DocTable takes it from there — the two are complementary, not alternatives.
Is there a free OCR option?
Genuinely free OCR tools do exist, and it is worth knowing what you get from them: almost all return plain text — a stream of characters with the table structure thrown away. That is fine for reading a page and useless for a spreadsheet.
DocTable is not free beyond a starter allowance. Signing in gives 10 free scans, after which each scan uses a paid credit; downloading a result never costs extra. The 10 scans exist so you can check the extraction quality on your own documents before paying anything, which is the part a free text-only tool cannot tell you.
Document AI, IDP, text extractor — the same job under several names
The vocabulary in this space is unusually crowded. Document AI, intelligent document processing (IDP), document data extraction and text extractor all describe some version of one task: take a page meant for human eyes and turn it into fields and rows a machine can use.
The practical split is not the name but the output. A text extractor gives you the words. Document-AI-style extraction gives you which words — vendor, date, total, this column, that row. DocTable does the second, and returns it as Excel, Word or CSV rather than as an API response, because the job it is built around ends in a file somebody opens.
Frequently asked questions
Is DocTable's AI OCR free?
Sign-in is required before the first scan (Google or verified email) — there's no anonymous scanning. Sign-in gives 10 free scans, then scans use paid credits. Downloading a result is free; the charge happens at scan time, not download time.
How is AI OCR different from classic OCR?
Classic OCR recognizes characters and returns raw text. DocTable's vision-language model reads structure — tables, label/value fields, stamps — and returns organized rows and fields instead.
Does it work on scanned PDFs and photos, not just digital PDFs?
Yes. Scans, phone photos and screenshots use the same vision OCR as digital PDFs. Very low resolution or heavy blur reduces accuracy.
Does it handle handwriting?
Clear block handwriting on forms is often readable. Cursive or messy handwriting is not reliable — always verify those fields manually.
Persian / Arabic / Chinese?
Yes. Same default OCR model for all languages, independent of the UI language.
Invoice OCR?
Yes — vendor, date, line items, tax and total are extracted as editable rows, ready for Excel.
Related pages
Temporary processing. Always verify OCR output before financial or legal use. Default OCR model is the same for every language. See Privacy in the app.