Lovelaice framing

Failure pattern

Definition

A recurring, nameable way your feature breaks. Patterns are fixable. Individual bad outputs are anecdotes — and anecdotes are what your team is currently trading in Slack threads.

A failure pattern is a specific, recurring way your AI produces bad output — named, described, and grounded in observed examples. Patterns are what error analysis produces and what improvements target. They're also what LLM-as-judge criteria should be written against, because judges need something concrete to detect.

Why it matters

You cannot fix what you cannot name. A team that talks about 'weird outputs' argues in circles; a team that talks about 'fabricates dates when the source is missing them' knows exactly what to fix, test, and monitor.