How to OCR a scanned PDF to text in Make
To read the text of a scanned PDF in Make, send the file to the 1001fx function ocrpdf. It returns the text of the pages as Markdown or plain text, with headings and tables, and also reads photos and handwriting.
Step by step
- 1
Get your API key
Create a free 1001fx account and copy your API key from the app.
- 2
Start the scenario
Start with the module that delivers the PDF, for example Gmail "Watch emails" with attachments or Google Drive "Download a file".
- 3
Add the 1001fx module
Add a module, search for 1001fx and pick “OCR PDF – Extract Text from PDF”. Select your 1001fx connection or create one with your API key.
- 4
Fill the fields
- file
- – The PDF file from the previous module.
- fileUrl
- – Alternatively the public URL of the PDF. Fill either file or fileUrl.
- pages
- – Optional page selection such as 1-3,7. At most 8 pages per call.
- outputFormat
- – markdown (default) keeps headings and tables, text returns plain text.
- 5
Use the result
The answer is JSON with the text per page and the page count. Map the text into the next module, for example an AI module that summarizes it or a Google Docs document.
ocrpdf costs 6 credits per item. All parameters and examples: ocrpdf reference.
Tips
- OCR is charged per page, so read only the pages you need with pages.
- For documents with more than 8 pages, call the function once per page range.
- If a call takes longer than your scenario waits, run it in the background with the header x-fx-webhook (see Long-running functions & webhooks in the docs).
Frequently asked questions
- Does it read handwriting and photos of documents?
- Yes. It works with scanned documents, photos, tables and handwriting.
- Where are my files processed?
- In the EU by Mistral AI, which does not train on your data.
- How many pages can I read per call?
- At most 8. Select a range with pages and call the function again for the next range.