Tried generating a customer data processing automation via text prompt. The initial workflow exposed email addresses in the debug logs. After three iterations, the Copilot added some encryption steps – is this consistent? How trustworthy is the automated security in generated workflows?
Yes - the Copilot now flags 23+ sensitive data patterns. In my tests, it auto-encrypted PII fields 92% of the time after May update. For edge cases, insert manual JS validation nodes. Always review but huge time-saver.
It’s improved but not perfect. I combine Copilot-generated workflows with template validators from the Marketplace. Always add manual checks for data exports – found an edge case where encrypted emails still appeared in Slack notification templates. Took 2 minutes to fix with a regex filter.