💬 How to build a workflow that detects new bullet points in a Google Doc (DE), translates them with ChatGPT, and appends them to another Doc (FR)?

Help me Build my Workflow
1

Hi everyone :wave:

I’m trying to build the following automated workflow in n8n and would love your help on how to best structure it:

:repeat:

Goal:

I have two Google Docs:

  1. Original Document (DE) → written in German
  2. Translated Document (FR) → should receive the French translations

Whenever I add new bullet-point-style lines (e.g. marketing headlines) to the original document, I want n8n to:

  • Detect only the new lines (not already processed)
  • Translate each new line to French using ChatGPT (OpenAI)
  • Append the translated version to the bottom of the translated document

:brain:

Current Setup:

  • I’m using Google Docs → Get to read the German document
  • I’m using a Static Data node to store the last known content (body field)
  • In a Function node, I compare the current vs. previous content (split by line)
  • I plan to pass each new line to OpenAI and then use Append Text for the French doc

:x:

Problem(s):

  • The Function node keeps throwing:Referenced node doesn’t exist…even though both nodes (Google Docs & Static Data) are connected
  • I’m unsure how to correctly pass two inputs into a Function node
  • I also want to update the Static Data after each successful run to store the new DE content

:bulb:

My question:

How would you structure this workflow cleanly in n8n?

Especially:

  • Best practices for comparing two text inputs line-by-line
  • How to pass two different inputs into a Function node
  • How to reliably store and update previous values between runs (via Static Data or better approach?)

I’m happy to share my JSON export if needed. Thank you in advance :pray: