Shopping agents put pressure on Amazon’s retail model

AI shopping agents are moving from theory toward consumer products, with Instinct and Meta’s Muse pitching bots that can buy groceries, book flights, and handle other transactions on a user’s behalf. The shift could reshape online retail by moving shopping decisions away from store websites and into third-party assistants, though consumer trust, reliability, and technical limits remain major obstacles.

Early efforts have exposed practical problems. OpenAI’s Instant Checkout allowed purchases inside ChatGPT, but Walmart later said conversion was poor because shoppers did not want to buy one item at a time and preferred building a cart. Real-time pricing and product data also remain difficult for AI systems, and some retailers are reluctant to make that information easy to access.

Amazon is especially exposed if agents become a common shopping interface. A successful agent could reduce Amazon to a fulfillment source while limiting its ability to recommend products, promote alternatives, or deepen customer engagement. Amazon has responded by limiting bot access, withholding up-to-date pricing data, suing Perplexity, and building AI features inside its own ecosystem.

The rise of agents also raises questions about incentives and advertising. If a shopping bot works for the consumer, it may need to prove that it has no preferred vendors, but free services could eventually rely on commissions, marketing, or transaction fees. Major disruption to advertising is viewed as unlikely in the next 24 months, though brands are already watching the shift closely.