Mahal Systems

Overview › 6Topologies › a.Escalating

Escalating

Escalate from cheap to capable models. 1 optimization.

  1. i.

    Oracle-worker multi-model

    Let a cheap model do the work and call an expensive one only when a trigger fires: a cascade escalates when the cheap answer looks unreliable or fails a check, a router picks the model before the call, and an advisor tool lets a cheap agent consult a stronger model mid-task.

    Better governanceMixed qualityMixed costMixed performance

    Tier 4 · restructuring · changes model · any agent · 6 sources, 5 measured

See also

  1. 9.c.iii

    Audit a verifier-gated cascade against a gold sample the verifier never sees

    When a cheap model's outputs are accepted or escalated by a verifier, measure true error on a small gold-labeled sample outside the verifier: metrics measured through the verifier showed about 3% error while true error was 14-32%.

    Verification › Sizing verification

    Better qualityHigher cost

    Tier 4 · restructuring · changes control flow, model · any agent · 13 sources, 13 measured

Before you go...

Build better agents, weekly

Our latest research, performance benchmarks, optimization strategies, and updates around AOT development.

One email a week, nothing else. Unsubscribe anytime.