Anthropic released "Building Commerce Agents with Claude" on September 2, a blueprint of software patterns and guardrails to help companies build AI agents for retail and commerce applications (Practical Ecommerce). The blueprint includes two examples: a consumer-facing shopping agent that resides on a store's website or app and connects to product data and checkout systems, and a merchant agent that works behind the scenes to recommend inventory actions and price changes (Practical Ecommerce).
Retailers running Anthropic's shopping agents have experienced a 35% increase in order value and a 60% improvement in conversion rate (Practical Ecommerce). The announcement arrives as consumer adoption of AI shopping platforms accelerates: 24% of U.S. online buyers plan to start their 2026 holiday shopping on AI platforms such as Claude, Google Gemini, and ChatGPT, up from 17% in 2025 (Practical Ecommerce). Two competing models are emerging—external platforms like ChatGPT owning the AI relationship and checkout, or merchants deploying their own on-site agents to retain customer relationships and brand control (Practical Ecommerce).
For commerce practitioners, the strategic implication is clear: merchants that know their customers, retain first-party data, and maintain direct channels such as email and loyalty programs should be better positioned regardless of whether shoppers arrive through an outside AI platform or use the merchant's own agent (Practical Ecommerce).