Anthropic has published blueprints and templates for building AI-powered shopping and merchant agents using its Claude models, aiming to help e-commerce businesses implement automated purchasing systems. The move comes despite surveys suggesting most consumers remain reluctant to let AI make purchasing decisions on their behalf, meaning Anthropic faces a significant task in persuading both shoppers and retailers to adopt the technology.

The blueprint, available via GitHub, offers reference implementations for retail, travel, telecom and ticketing platforms, letting agents connect to product catalogues, checkout systems and purchase history. A Gartner survey found only 11% of consumers would let AI make purchase decisions, though an Accenture poll was more upbeat, finding 32% open to handing over purchasing and 9% willing to accept fully autonomous shopping. Critics have raised concerns that such agents could worsen dynamic pricing, with Groundwork Collaborative noting Walmart's "Sparky" AI assistant users already spend 35% more, while Chargebacks911's CEO warned of unresolved fraud liability when consumers dispute AI-made purchases.