Power Automate Integration

Convert PDFs to structured data inside Microsoft Power Automate using the PDFPipe API. This guide walks through the complete setup using the built-in HTTP action - no custom connector required.

Prerequisites: A PDFPipe account with an API key and a Power Automate Premium or Per-User licence (required for the HTTP action).

Create a new flow

Open Power Automate and create a new Automated cloud flow or Instant cloud flow. Choose a trigger that provides a PDF URL, for example:

SharePoint - "When a file is created or modified", then a Create sharing link action set to "Anyone with the link" so the URL is publicly fetchable
Outlook - "When a new email arrives" where the email body contains a public PDF link
Manual trigger - for testing, with a public PDF URL as input

The URL must be reachable without a login. PDFPipe fetches the PDF from its own servers and does not send your Microsoft 365 cookies or credentials, so a tenant-only SharePoint or OneDrive link will fail with URL_UNAUTHORIZED, URL_FORBIDDEN, or REDIRECT_TO_LOGIN. Use an anonymous sharing link, a public URL, or a link with its access token in the URL. Email attachments themselves are not URLs and cannot be sent to PDFPipe directly.

Add an HTTP action to call PDFPipe

Add a new step and search for the HTTP action (built-in). Configure it:

Method: POST
URI: https://api.pdfpipe.dev/v1/convert
Headers:
Authorization: Bearer pk_your_key_here
Content-Type: application/json

Set the Body to:

HTTP action body

{
  "url": "@{triggerOutputs()?['body/Link']}",
  "format": "json",
  "returnMethod": "inline"
}

Replace the dynamic expression with the URL from your trigger. Use "format": "text" or "markdown" if you prefer a different output. Remove "returnMethod": "inline" to use the default presigned resultUrl instead.

Handle the response

For small PDFs (<10 MB), you'll get a synchronous 200. With returnMethod: "inline", the body has content (a string) and contentType. With the default file delivery, use resultUrl. Add a Parse JSON action after the HTTP step to extract the fields; for format: "json" add a second Parse JSON on content (using json(body('Parse_JSON')?['content'])) to reach pages[].text, metadata, and totalPages.

For larger files, the response is 202 with a pollUrl. Add a Do Until loop:

Inside the loop, add an HTTP GET action to https://api.pdfpipe.dev@{body('Convert_PDF')?['pollUrl']} (append ?returnMethod=inline to include content when complete)
Set the loop condition to continue until status equals complete
Add a Delay of 5 seconds between iterations

Do Until condition expression

@equals(body('Poll_Status')?['status'], 'complete')

Use the result

Once the status is complete, use the resultUrl in a follow-up HTTP GET action to download the converted data. From there you can:

Save the JSON or text output to SharePoint or OneDrive
Send extracted text via Teams or Outlook
Insert one row per page into Excel or Dataverse
Pass the page text to AI Builder or Copilot to pull out fields like totals or dates (PDFPipe itself returns text, not fields)

Error handling (recommended)

Wrap your HTTP actions in a Scope action and add a parallel branch with "Configure run after" set to "has failed". This lets you log errors or send notifications when a conversion fails.

PDFPipe returns structured error codes (see the error reference) that you can parse to determine whether to retry or alert.

Tips

Store your API key in a Power Automate Environment Variable or Azure Key Vault rather than hard-coding it.
For batch processing multiple PDFs, see the Batch Processing guide.
Use "format": "markdown" when feeding results to AI Builder or GPT-based Copilot actions.