An explanation, drawing on Kathy Sierra's "identity as superhero" idea, for why automated code-review bots historically flopped while agentic coding tools like Claude Code and Codex succeeded despite doing conceptually similar work. Review bots publicly critique the developer's own code — an attack on identity and status. Agentic tools instead brand the output as "you + the tool," letting the human keep the bragging rights of being a good developer while the tool does the labor.
The difference isn't the intelligence of the automation — see Intelligence Is Not the Bottleneck — but whether the interaction preserves or threatens the user's sense of competence and identity.
When designing any tool that evaluates or corrects a user's work, frame it so credit still accrues to the human (collaborator framing) rather than exposing them as having been wrong (critic framing).