When tier-B cost confidence is enough.

Not every action needs the same certainty about cost. Re-ranking a product in search is reversible and cheap to get slightly wrong. Cutting its price is neither. That asymmetry is why Magistry gates actions by cost-confidence tier — and why we just unlocked OPTIMIZE_LOSER for Tier B.

The three tiers

• Tier A — landed cost is known and current. The full action set is available, including price moves.
• Tier B — cost is estimated within a confident band, but not exact. Soft, reversible actions only.
• Tier C — cost is unknown or stale. The agent can observe and propose, but not act on margin-sensitive moves.

The tier is stamped on the decision row at the moment of action, so an auditor always sees the confidence the agent actually had — not what we know now.

Why OPTIMIZE_LOSER is safe at Tier B

OPTIMIZE_LOSER doesn't touch price. It demotes a chronic underperformer in merchandising — less promotion, lower placement — and it's fully reversible with a stored op. The downside of being wrong is small and recoverable: a SKU gets less visibility for a while, and one row reverses it. That's exactly the risk profile Tier B is meant for.

{
  "product_id": "sku_2218",
  "action": "OPTIMIZE_LOSER",
  "cost_confidence": "B",
  "effect": "demote_merchandising",
  "touches_price": false,
  "reverse_op": "RESTORE_RANK sku_2218",
  "decision": "approved"
}

Contrast that with PRICE_DROP, which stays locked until Tier A. A wrong price on an underestimated cost is a margin leak you might not catch for weeks, and reversing it doesn't refund the orders that already sold cheap. Same SKU, same uncertainty — different blast radius, different gate.

The trick isn't being certain about everything. It's matching the reversibility of an action to the confidence you have. Soft and reversible can move on estimates; expensive and sticky has to wait for facts.

— Ines Park

Tiering cost confidence is how an agent stays useful when its information is imperfect — which is always. It acts where being wrong is cheap, and it waits where being wrong is expensive. That's not timidity. It's the whole discipline.