Describe the problem/error/question
What is the error message (if any)?
Please share your workflow
(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
Share the output returned by the last node
Hi everyone,
I’m building a workflow in n8n to process bank statement PDFs using Google’s services.
Workflow steps:
Gmail → receive PDF
Upload to GCS
Call DocAI batchProcess with a bank statement processor
Get JSON output from GCS → parse → send to AI Agent
Problem 1 – Page limit
With Document AI, whenever I send a PDF larger than 30 pages, only the first 30 pages are processed. The rest of the file is ignored.
Problem 2 – Wrong numbers
Even inside the processed pages, the numbers are often incorrect (totals, balances, or transaction amounts don’t match the original PDF).
Extra test with Google Vision OCR
I also tried Google Vision OCR (files:asyncBatchAnnotate) directly:
It processed 100+ pages without problem 
But the results are very poor 
(text scattered, numbers often wrong, tables not aligned).
Environment:
n8n v1.107.3 (self-hosted)
HTTP Request v3 node calling batchProcess
Google Document AI processor type: bank statement
Auth via service account (already works fine for GCS)
Questions for the community:
- Is the 30-page limit for Document AI expected? Any way to process larger PDFs in n8n?
- Why might DocAI be producing wrong numbers (inconsistent with original PDF)? Are there recommended processor settings or strategies (e.g., splitting input, using field masks)?
- Since Vision OCR handles bigger files but with poor quality, is there a best practice to combine or enhance both approaches?
- For large PDFs, what’s the recommended way in n8n to split the file into chunks and merge the results?
Thanks a lot
Information on your n8n setup
- n8n version:
- Database (default: SQLite):
- n8n EXECUTIONS_PROCESS setting (default: own, main):
- Running n8n via (Docker, npm, n8n cloud, desktop app):
- Operating system: