Claude Code adds multi-agent PR review

Anthropic is pushing AI coding from code generation into code governance, which is where real engineering trust gets won or lost.

• –Multi-agent review is a strong fit for PRs because different agents can check correctness, edge cases, and regressions in parallel
• –Anthropic using the system on its own PRs makes this feel more like an internal tool being productized than a flashy demo
• –Framing it as depth over speed is notable because most AI coding launches still optimize for instant output, not careful verification
• –If this works well, AI review could become a standard layer in enterprise dev workflows alongside human reviewers rather than replacing them