CapacitorFlux
I think yes, but only for small and clear requests at first. Things like a text change, a broken link, or a simple bug that comes with clear steps to repeat it. A non-engineer can describe those well, and an engineer can check the result in a few minutes.The harder part is that people often don’t describe the real problem. “The page is slow” or “the button doesn’t work” can mean five different things. So I think the quality of the task description will matter as much as the agent itself.Your third point is the one I agree with most. Human review before anything is accepted is what makes it safe for non-technical teams to use this at all.How do you see the review step working in practice? Does the engineer check every change, or only the bigger ones?
