Hi Amit K., Thank you for investigating. I have enabled auto-update and re-run the column as instructed. However, this does not appear to be a stale data issue. Here is what we are observing:
The column is executing and generating new errors, not displaying old data.
Every run consumes API credits and produces fresh error messages with new timestamps: "Unable to retrieve current employment information from LinkedIn" and "Unable to access LinkedIn profile for current employment details." Stale data would show cached results without credit consumption. We are seeing active processing that completes with failure states.
The auto-update setting does not explain this failure.
The Agentiiv column is in the same table as three other client columns with identical settings. Those columns process the exact same LinkedIn URLs successfully. If auto-update were the cause, all columns would fail identically. We have also rebuilt the Agentiiv column from scratch in separate workbooks with default settings. Each fresh build failed immediately on first run before any settings could become stale.
This failure started after Clay's platform incident few days back.
This column worked reliably for seven months with the current settings. The failure began immediately after the documented platform incident and has persisted across multiple complete rebuilds, different AI models including GPT-4o and Claude, different workbooks and tables, and with auto-update both enabled and disabled.
What we need from your team:
This is now a critical issue for our operations. We need your engineering team to examine the backend execution logs for this column and identify any job routing differences between Agentiiv and our working client columns, or any changes that occurred during the incident. The failure pattern suggests this is a platform-level issue rather than a configuration problem on our end. We have exhausted all user-level troubleshooting and require backend diagnostics to move forward. I am attaching a screenshot showing the current state. We need this resolved or require a credit refund for the testing attempts, as this is blocking our client deliverables.
