Public teardown library
Every public result keeps the joke attached to a useful buyer artifact: claim, hidden assumptions, one main badge, supporting risks, and one-week test. Newest verdict first.
Filter by vendor
Anthropic
Anthropic Claude Code
“A changelog is no substitute for showing how AI-driven outbound truly juggles messy CRM fields and routing rules.”
Claude Code updates sound impressive but leave GTM teams guessing on CRM integration, lead routing, and meeting qualification impacts.
Support / product assistant
Anthropic Claude Code“AI says 'autonomous' but your AEs still review every PR and handle merge conflicts.”
Claude Code automates multi-file code changes and test runs locally but depends on user approvals and existing dev workflows.
Hidden assumptions
User has permission and technical ability to install local tooling · User consistently approves or rejects AI-suggested changes
