Dataleo Insight · 2026-06-16· Supply Chain AI
PepsiCo documents AI-driven pricing and promotion optimization at scale
A new arXiv paper details how PromoAI and PricingAI combine forecasts, elasticity models and optimization constraints in commercial planning.
PepsiCo has documented two deployed decision systems for commercial planning: PromoAI, focused on promotion-calendar optimization, and PricingAI, focused on base-price optimization. The paper describes a stack combining machine-learning forecasts, elasticity modeling, mixed-integer or nonlinear optimization and operational business constraints.
For Dataleo, the signal is not simply that AI is used in pricing. It is that commercial decisions are being formalized as optimization workflows: forecasts, financial targets, retailer economics and execution constraints are treated as part of the same decision system.
This matters for Supply Chain Planning, Demand Planning and Revenue Growth Management teams because pricing and promotion choices reshape demand, inventory exposure and replenishment pressure. The main condition for value is explainability at the business-rule level: planners and commercial teams must understand which constraint, elasticity assumption or objective function is driving the recommendation. The failure mode is a mathematically strong optimizer that loses adoption because teams cannot challenge or override its logic during real planning cycles.
For Dataleo, the signal is not simply that AI is used in pricing. It is that commercial decisions are being formalized as optimization workflows: forecasts, financial targets, retailer economics and execution constraints are treated as part of the same decision system.
This matters for Supply Chain Planning, Demand Planning and Revenue Growth Management teams because pricing and promotion choices reshape demand, inventory exposure and replenishment pressure. The main condition for value is explainability at the business-rule level: planners and commercial teams must understand which constraint, elasticity assumption or objective function is driving the recommendation. The failure mode is a mathematically strong optimizer that loses adoption because teams cannot challenge or override its logic during real planning cycles.
Keywords