Prompt routing: when to use Claude vs GPT vs Gemini in the same agent

Model selection is the single biggest lever nobody talks about when building reliable agents for real client work.

Most agent failures are not prompt failures. They are routing failures.

You ask one model to do everything, then blame the prompt when the job needed different kinds of intelligence at different moments.

The pattern

Use the model that fits the step:

The rule

Do not route by brand preference. Route by job shape.

That's it for this week. If this letter was useful, the best thing you can do is forward it to one person who'd get value from it.

Reply to tell me what you're working on. I read everything.

- Hichem

Not a subscriber yet?

Get the next letter in your inbox every Tuesday. Free. No nonsense.

No spam. One click unsubscribe.