@Joseph_Iko1 usually this is the trigger feeding different data than your manual test, by hand you run on pinned or sample input but the live trigger sends a different shape (or empty), so an expression that resolved manually comes back undefined and the node throws. open the failed run in the Executions tab and check the exact error and which node fails, whats it say?
The incoming data may be different during automatic runs. Compare the execution data from both runs and check that all credentials and environment variables are available.
One specific thing to add: click “Listen for Test Event” on your trigger node, then trigger it once with a real event - this lets you inspect the exact data structure the trigger receives in production. That data is almost always different from what manual test mode gives you.
Then check the Executions tab, find the failed run, and click the failing node - it shows exactly which expression broke and what the actual input data looked like at that moment. The mismatch is usually either missing fields or data nested differently than expected.