Effort Routing

Simple Definition

Effort routing is choosing how much reasoning depth, context, persistence, tool access, and review an AI task receives, based on how complex it is and how costly it would be to get wrong.

Effort Routing vs. Model Routing

These are different axes. Model routing decides which model handles a task. Effort routing decides how hard that model or agent works on this specific task, quick and light, or deep with full verification. You can route a task to the same model at two different effort levels and get very different amounts of context, checking, and review.

The Rough Tiers

• Quick: fast, minimal context, no formal review
• Standard: normal reasoning, relevant context, a light self-check
• Deep: more context, a plan before execution, an explicit review step
• Critical: full context, a verification gate, and required human approval

Why It Matters

Using critical-mode effort on every task wastes time and money on things that don’t need it. Using quick-mode effort on an irreversible, high-stakes action under-protects the work that actually matters. Effort routing makes that tradeoff explicit instead of leaving it to whatever a tool defaults to.

Example

A team drafting five social post variations uses quick mode, no review needed since picking a different variation costs nothing. The same team editing authentication code uses critical mode: full context, a plan, tests, a diff review, and a human sign-off before it ships.

Related Terms

• Model Routing, choosing which model handles a task
• Task Budget, the expected cost of running a task at a given effort level
• Verification Gate, the check that deep and critical tasks require
• Human-in-the-Loop, the approval step for critical-effort tasks

For a full framework, see Match the AI Agent to the Risk.