AI Agent Effort Prioritization Framework A three-step framework for selecting AI agent projects across value lenses, scoring axes, product-line questions, and repeated review. AI Agent Effort Prioritization Framework HandiProducts family only - value lenses, evidence-based scoring, per-line judgment, and regular re-scoring 1 VALUE LENSES Map the opportunity to a real business lever Revenue Conversion rate, AOV, close rate, retention, and recovered abandoned carts Production Throughput More units or orders processed without adding headcount or scheduling friction Customer Service Cost Minutes, calls, escalations, returns, and WISMO tickets removed from people A candidate must name at least one lens explicitly. "Helps operations" is not enough. 2 SCORE EVERY CANDIDATE, 1-5 EACH Use the same five axes every time Dollar / Time Magnitude How big is the lever? Evidence Quality Data, or just a guess? Build Readiness Existing integration? Blast Radius / Risk Customer-facing? Time-to- Value Days or weeks? Fastest wins = high magnitude + high evidence + high readiness. Novelty does not get bonus points. 3 PER-PRODUCT-LINE DISCUSSION Ask this for each brand, not once for the company Dominant constraint Demand-capped, capacity-capped, service-cost-capped, or mixed? Sales motion One-time sale, recurring/repeat motion, or pipeline compounder? Evidence source Point to actual data before assigning the lens. Integration foundation If the data plumbing is missing, mark readiness down first. Prioritize what wins on multiple axes High magnitude + high evidence + high readiness beats anything net-new, no matter how exciting. Re-score as bottlenecks, data, or context change