Unified shopping intent routing
Stuart Kearney, co-founder of Vetted, on AI agents in shopping
The winning shopping interface will route between search and conversation inside one box, because shoppers switch constantly between finding a known item fast and getting help on an unclear decision. Amazon already separates those jobs today, with Rufus living beside the core search bar, but the logic points toward one front door that detects intent and then changes the experience underneath, while keeping Amazon’s speed, reviews, shipping promises, and familiar product pages intact.
-
Amazon has kept Rufus as a separate assistant while scaling it broadly. Rufus became available to all U.S. customers on app and desktop in 2024, and Amazon said in its Q3 2025 earnings that 250 million customers had used it that year. That supports the idea that Amazon is training usage patterns before folding AI deeper into core search.
-
The split maps to two very different jobs. For navigational shopping, users often know the SKU shape they want and just need filters, ranking, shipping speed, and reviews. For research queries, users need back and forth clarification, comparison, and explanation. Vetted describes strong multi turn behavior for those messy decisions, especially in long tail categories.
-
Google is moving in the same direction. In 2025 it brought conversational shopping, rich product results, price tracking, and agentic checkout into Search and AI Mode, showing that large platforms are not building a separate AI mall. They are grafting AI help into the existing shopping surface where intent already starts.
The next step is intent routing becoming invisible. The search bar stays, but behind it the system decides when to return a fast ranked list, when to open a guided comparison flow, and when to take action like tracking price or checking out. That makes unified shopping interfaces more useful without asking consumers to learn a new behavior.