Anthropic lays groundwork for bots that shop for you

Anthropic has developed a framework for creating shopping agents powered by its Claude AI. While consumers are currently hesitant to fully entrust AI with purchase decisions, the company has released a set of templates and reference implementations to encourage e-commerce businesses to adopt automated purchasing agents. These agents can understand natural language queries, compare products, and complete transactions for users.

The blueprint encompasses the necessary components for an engineering team to deploy a commerce agent in a matter of days. It includes reference implementations for shopping agents and merchant agents tailored to various industries, such as retail, travel, telecom, and ticketing platforms. The agents can interface with product catalogs, shopping carts, checkout systems, and databases containing customer preferences and purchase histories.

A key feature of these agents is their built-in constraints to prevent excessive price fluctuations and manipulative upselling. By limiting prices to catalog data and avoiding aggressive upsell tactics, the agents aim to maintain fairness in pricing. However, concerns persist regarding the potential for AI agents to exploit dynamic pricing, where individual customers are charged different prices based on their behavior and historical data.

Consumer sentiment towards AI-powered shopping remains mixed. According to a recent Gartner survey, only 11 percent of respondents are willing to allow AI to make purchase decisions. However, a larger Accenture survey indicates that three-quarters of consumers would permit AI agents to handle routine tasks, and a third are open to fully autonomous shopping with AI. Despite these optimistic findings, Anthropic acknowledges that the biggest hurdle to widespread adoption is building consumer and merchant trust in AI agents.