I’ve been reading about AI copilot workflow generation—the idea that you describe what you want in plain text and it spits out a ready-to-run workflow. It sounds incredible on paper, but I’m skeptical.
Every tool I’ve used that promises “describe it in English and we’ll build it” ends up delivering something that’s 70% there. Then you spend twice as long editing as you would have building it from scratch. I’m wondering if this is a real timesaver or just shifting the cognitive load from building to debugging.
Has anyone actually used this to push something to production without significant rework? I’m specifically curious about workflows that handle business logic, error handling, and edge cases. Not just simple “send email when spreadsheet updates” stuff.
What’s the realistic turnaround? From plain text description to deployed automation—how much iteration actually happens?