Automated code review bots slowing down PR cycles?

We've been running automated code review bots (lint, security, style checks) on every PR and they've started to bottleneck our merge velocity. A medium PR now takes 20+ minutes just for CI to clear before humans even look at it. How did your teams balance thorough automated checks against PR cycle time? Did you move to parallel pipelines, selective triggers, or just accept the wait? Stack: GitHub Actions, Python/TypeScript repos, ~40 devs.