Hi there I was hoping to use previous automation runs as a sort of "poor-man's training set." Since I'm already saving the results of each run in a Google Sheet, I could tap into that data to drive dynamic prompt enhancements in new workflows.

I was thinking along the lines of:

Use the stored run results from the sheet, categorizing them as good or bad examples.

When a workflow starts, pull in relevant examples from the sheet to give some extra context. This helps steer the process based on what’s worked (or not) in the past.

It's the only way I can think of to actually enable any type of reinforcement learning or trying to improve and classification type tasks.