1
Took a harsh critique from a senior dev about my AI prompt chaining and it totally flipped my approach
He said my prompts were too vague and I needed to break each step into its own atomic task instead of cramming everything into one big block. Now I'm getting way cleaner outputs after 3 days of rewriting my workflows - has anyone else had to redo their whole setup after someone pointed out the obvious?
3 comments
Log in to join the discussion
Log In3 Comments
alicehernandez1d ago
Yeah, the "you're trying to do 5 things at once" part really hits home. I went through the same thing last week with a chatbot workflow where I had it summarize, extract data, and rewrite tone all in one go and it came out a mess. Once I split those into separate steps like "pull out the numbers" then "rephrase it friendly," it actually started making sense.
4
the_grace1d ago
Oh man, I had almost the exact same thing happen about 2 months ago. A senior dev named Mike sat me down and basically told me my prompts were just one giant paragraph that tried to do 5 things at once. I felt like an idiot for not seeing it myself. After splitting everything into tiny steps like "identify the main issue" then "suggest three fixes" instead of just "fix this," my outputs got way more consistent. It was annoying to redo all 12 of my workflows but now I actually trust what comes out.
2