Routing Simulator
Answers the question: why did this task go to this model, under this policy, with this review path?
What this shows
Budget, sensitivity, destination and mandate can shape routing in inspectable ways. A firm can separate raw generation from the policy that chooses path, so “which model” is not the whole story.
Where this would sit in a firm
AI steering groups, architecture reviews and vendor comparisons where teams need explicit rules, not a black box that picks an endpoint.
Demo only
Routes are chosen by fixed demo rules, not by calling external models or your firm’s configuration. Not legal advice.
What the rules illustrate
- A low-risk internal summary and a client-facing advice note should not follow the same path, even when the prompt looks similar.
- Routing decisions should be explicit enough to inspect: model tier, governance and flags, not a black box.
- Budget pressure and sensitivity can trade off only when policy allows it.