Help with Integrating Azure Document Intelligence API After Form Submission

Help me Build my Workflow
1

Hi everyone,

I’m working on an automation in n8n where a user submits a form with the following fields:

  • Name
  • Surname
  • ID Number
  • ID Document (uploaded file)
  • Proof of Address (uploaded file)
  • Selfie ID (uploaded file)

After the form is submitted, I want to send the uploaded ID Document to Azure Document Intelligence API using the prebuilt-idDocument model to extract key data like the full name, ID number and DOB.

I tested the API in Postman , and it returns a raw JSON response successfully — but it’s not structured with clear fields, just a general content field. I was hoping to extract the fields (e.g., Name, DOB, ID Number) directly in n8n , but:

  • When I send the same request via HTTP Request node in n8n, it does not return structured data or fails entirely .
  • Even when trying to replicate the Postman setup, it doesn’t seem to work in n8n.
  • The API expects a POST request with a binary file (PDF/image), and I believe the issue might be around sending the file properly from the form upload to Azure via the HTTP node .

Has anyone successfully connected Azure’s Document Intelligence with n8n for ID extraction? Thanks in advance!

2

Hi, how did you send the PDF documents to Azure Documents Intelligence?

3

Hello
Were you able yl resolve the same ?!

Regards,

4

Not a solution to your problem but alternative, you can use Azure DI SDK (via code and langchain node in n8n) OR REST API, in my case i used code node (javascript) + REST API rather then using HTTP node, i followed Leverage Azure Document Intelligence in Langflow and n8n

5

Hi Vito246,

I see you have the raw JSON coming back from Azure’s prebuilt-idDocument model in Postman but hit issues when you send the file from n8n. Here’s what’s possible: with a single HTTP Request node and a short Code node you can get structured fields (Name, DOB, ID) straight after the form submits.

  1. Make sure the HTTP node sends the file as binary:
  2. • Set “Send Binary Data” to true.
  3. • In the body parameters choose the binary property that stores your uploaded PDF/image (e.g. document).
  4. • Add the header Content-Type: application/pdf or the correct mime-type.
    1. Azure expects the API-key header plus the model URL:
  6. • URL: https://{region}.api.cognitive.microsoft.com/documentintelligence/documentModels/prebuilt-idDocument/analyze?api-version=2023-10-31
  7. • Headers: Ocp-Apim-Subscription-Key: {{ yourKey }} and x-ms-useragent: n8n.
    1. The response comes back as an async operation. Grab the operation-location header, wait 2-3 seconds, then do a second HTTP node (GET) to that URL. That call returns the JSON with documents[0].fields where Name, IDNumber, DateOfBirth live.
    1. Use a Set or Code node to map those fields:
  10. `{ name: item.json.documents[0].fields.customerName.content,
  11.   idNumber: item.json.documents[0].fields.documentNumber.content,

  12.   dob: item.json.documents[0].fields.dateOfBirth.content }`
    1. Finally push the parsed data wherever you need (DB, email, etc.).
      In similar KYC workflows we cut manual data entry by 80 % and processed hundreds of IDs per hour by chaining those two HTTP nodes with a simple Wait.

The key is enabling Send Binary Data and doing the follow-up GET to the operation URL – that’s the piece Postman hides but n8n needs explicitly.

Have you already tried the two-step (POST ➜ GET) pattern, or adjusting the mime-type header?

Based on my experience – consult experts for specific advice. Hope this unblocks your ID extraction!

Cheers