Enterprise product managers must continually decide which features, initiatives and fixes deserve scarce engineering capacity. Prevailing scoring methods such as RICE, WSJF and MoSCoW depend on subjective estimates, are easily dominated by internal opinion, and treat customer evidence and business value as two separate conversations. This paper proposes the Integrated Customer-Value Prioritization Framework (ICVPF), a five-layer architecture in which AI competencies (infrastructure, business spanning and proactive stance) are converted into three product-management capabilities (information management, planning and implementation) that ultimately support customer lifetime value (CLV). ICVPF mines multichannel customer feedback with language models to produce a calibrated Customer Insight Score (CIS), estimates a Business Value Score (BVS) from CLV uplift, strategic alignment and risk, fuses the two through a learned composite value estimate, and allocates capacity with a constrained knapsack optimization wrapped in human oversight and explainability controls. Seven hypotheses (H1–H7) are derived from the conceptual model for future field validation. As a first test of the decision logic, we ran a Monte Carlo simulation of 1,000 planning cycles with 120 backlog items each. The integrated model captured 95.3% of the oracle value against 87.1% for manual RICE-style scoring (a 9.4% relative gain; paired t-test, p < 0.001), and its rank correlation with realized value was 0.872 versus 0.643. These results are synthetic: they demonstrate the internal logic of the framework, not field performance. We close with governance guidance and a research agenda.