Tier × effort routing from runtime signals

The idea

Each call's model tier and effort are chosen separately. The inputs are the stage, the risk of the pending action, the change size, open findings, failures, repository size and the budget left. Floors tune from verified outcomes.

The problem
One fixed model overpays on easy steps and is too weak for hard ones.
Why it matters
Model strength and effort follow each step's difficulty, risk and budget.
Where it stands
In production as of

A step-by-step replay of this idea is not published yet. The library adds one when a recorded run shows it.

Where it lives in the product

  • One effort allocation per item and stage

    One decision per item and stage picks low, medium or high effort. That level sets the stage's turns, time, cost, rounds and consultations, and the worker can ask for more.

  • Model choice controls and managed defaults

    Owners pick models per tier or fix the chains. People set floors or pins down to one step. Managed defaults move to new compatible models, and explicit pins stay fixed.

  • Model failure handling and outage recovery

    An unusable answer moves one tier up in the same call. Quota or auth failures put an arm on cooldown. A timeout retries at lower effort. A transient outage parks the item on a backoff timer and resumes the same run.

  • System One decision cascade

    Bounded choices (route, next action, failure class, done, risk, intent) are typed questions. A rule answers first, then a typed arm, then a small LLM, each with a confidence.

  • Engineer pipeline with stage shares

    Work follows an engineer's pipeline, and each stage gets a share of the time and cost. Review is one plan review and one holistic diff review, within about 10% of spend.

Turn the numbers into shipped work.

Crewshal runs these choices for you: a persistent AI worker with memory and rules, on your Claude and Codex subscriptions.