#contrarian

Ten ways AI customer-support deployments destroy CSAT — over-deflection, mis-routed escalations, transcript drift, brand voice collapse, more.

Eight observability anti-patterns that make agent traces useless — PII in spans, cardinality explosion, trace truncation, missing replay, more.

Ten AI-assisted coding anti-patterns that turn productivity gains into tech debt — accept-without-read, copy-paste sprawl, agent-blame, eval-skip.

Eight AI content pipeline anti-patterns that destroy quality at scale — stat fabrication, mid-doc factual drift, SEO over-fit, voice collapse.

Eight agentic workflow anti-patterns that turn agents into outages — hidden state, race conditions, un-bounded loops, tool over-broad scope.

Ten prompt engineering anti-patterns that quietly destroy production quality — few-shot pollution, instruction-stacking, format-via-example, more.

Seven RAG anti-patterns that destroy retrieval quality — chunk-too-big, retrieve-too-few, ignore re-ranking, no provenance, more failure modes.

Seven MCP server anti-patterns that turn agent tools into liabilities — tool-schema over-fitting, auth-after-build, chatty tool-call protocols.

Eight Claude Code team-adoption anti-patterns that stall rollouts — skill sprawl, CLAUDE.md bloat, permission drift, hook spam, model-version drift.

Ten agentic AI deployment anti-patterns — over-eager prod cut-over, missing eval gates, tool-call chaos, governance theatre, and seven more.