I’ve created a content repurposing workflow that takes text from a “Latest post to convert” column and uses AI to generate content for Twitter, LinkedIn posts, and LinkedIn hooks. The workflow works correctly when processing new content, but I’m experiencing an issue with filtering.
Current Setup:
Google Sheets Reader reads from my spreadsheet
Filter node checks if “LinkedIn Post” column is empty using “Is empty” condition
If empty, it passes the “Latest post to convert” content to three AI nodes
Each AI node (using Ask AI with Error Shield) generates content for different platforms
Google Sheets Writer nodes write back to respective columns
The Issue:
If I run the workflow without adding new content, nothing happens (correct behavior)
If I add a single new row to “Latest post to convert”, the workflow runs but ALSO re-processes ALL previous rows that already have content in the output columns
This is causing unnecessary credit usage as my AI nodes (using GPT-4.1-mini) are regenerating content for rows that are already complete.
What I’ve Tried:
Confirmed Filter node is set to “Is empty” condition
Verified connections between nodes
Tried different AI models
Questions:
Why is my Filter node not preventing already-processed rows from being re-processed?
Is there a better way to implement “process only new/empty rows” logic?
How can I ensure my workflow only spends credits on new content?
Hey @nicolas! If you’re reporting an issue with a flow or an error in a run, please include the run link and make sure it’s shareable so we can take a look.
Find your run link on the history page. Format: https://www.gumloop.com/pipeline?run_id={{your_run_id}}&workbook_id={{workbook_id}}
Make it shareable by clicking “Share” → ‘Anyone with the link can view’ in the top-left corner of the flow screen.
Provide details about the issue—more context helps us troubleshoot faster.
Hey @nicolas! Appreciate the detailed context. The issue here is that you have a separate Google Sheet Reader node downstream of your flow which outputs all the rows (doesn’t go through the filter), and it seems unnecessary in this case. You can avoid that by simply using the response from the AI in the last step.
Thank you! Super helpful. It works now. I’m unsure if I could remove the multiple writer nodes because some AI prompts depend on results that the previous AI step would give. I wasted a bunch of credits running those unnecessary rows I described on the post so might check later. What do you think?
I think you could still remove multiple writer nodes by just connecting the output of the previous nodes (I showed this briefly in the Loom) but either approach should be fine.