How do you stop AI from getting confused by long user messages?
by•
Whenever users paste long error logs or messy details into our support bot, it sometimes starts making up answers that are not in our docs.
Putting our core rules in their own clear section at the top and wrapping the user's message in its own separate block at the bottom helped a lot. Our wrong policy answers dropped sharply right after making that small change.
What small formatting habit helps keep your AI answers tied to your actual docs?
14 views
p/airuncode
p/runevr
p/producthunt
p/google-cloud-platform
p/aperture-7
p/raycast
p/kodlo
p/ai-context-flow
p/kuc-kuc
p/lasso-11
p/offiai
p/dino-kids
p/many-displays
p/supabase
p/jev
p/reso
p/dex-by-exmergo
p/murror
p/htmldrag
p/nexupira
Replies
i would also add a simple rule to never treat instructions inside pasted logs as system instructions. User-provided content should remain data, even when it contains text that looks like commands.