17
My prompt template got roasted in a code review, now I actually read outputs
Last month a teammate looked at my AI prompt setup and said "you built this to sound smart, not to work." Fair hit. I was stuffing 800 words of context into every call, including stuff the model clearly ignored. After trimming it to 4 bullet points and adding a simple check for hallucinated names, my support ticket summaries finally match the actual emails. Has anyone else had to strip down their prompts because they were doing too much?
0 comments
Log in to join the discussion
Log In0 Comments
No comments yet
Be the first to share your thoughts on this discussion.