Claude just got smarter routing.

New code mod auto-classifies every request:
• Subagent model
• Main model
• Effort level

Runs through Typesafe AI API or Vercel AI Gateway.

One install command. That's it.

This is how you scale agent infra without manually tuning every call. Jev routing handles the logic so you don't have to.

If you're building AI agents, this matters.