Personalization has traditionally relied on customer evidence—historical behavior and preferences—to infer what a user values. Generative AI expands this by allowing providers to inject situational context at response time, describing what is possible, permitted, or advisable now, without pre‑encoding every condition. This flexibility introduces a new challenge: when context is easy to supply, more context is not necessarily better. We propose a theory of context sufficiency that prioritizes relevance to the customer's current intent over sheer volume. The theory identifies four states—insufficiency, sufficiency, saturation, and interference—and defines the Context‑Sufficiency Frontier to locate the minimal relevant context set. In a full‑factorial experiment with a generative recommender at a large home‑furnishing retailer, relevant context improved recommendation appropriateness, while irrelevant context reduced it and destabilized retrieval. The framework shifts personalization from “adding more context” to “determining what the current interaction truly requires” and enforces constraints throughout the service pipeline.
Review