AgentMall is a commerce API for developers building personal AI agents that shop across major retailers. Instead of having an agent navigate store websites through a browser, developers connect it through REST or MCP to one service for searching live catalogs, assembling baskets, checking out, and receiving order updates. The service says it supports hundreds of merchants, with examples including Walmart, Amazon, Target, Costco, Instacart, Kroger, and Best Buy; its checkout flows are currently live at nine stores. Shoppers can describe purchases in ordinary language, compare available products and prices, and follow orders through confirmation and pickup or delivery.
The central argument is that checkout needs dedicated, reliable infrastructure rather than a general-purpose model clicking through web pages. Each purchase requires approval tied to the exact basket and a spending ceiling; if the total changes, approval must be renewed. Merchant credentials and customer profiles stay in a vault rather than being exposed to the model, and a virtual card can be assigned to each agent. AgentMall uses merchant-specific checkout flows to address bot protections, and says it checks with a merchant when an order submission is uncertain instead of blindly retrying and risking a duplicate charge. It also handles sign-in codes by asking the shopper, returns receipts and order status to the agent, and offers developers affiliate revenue on purchases.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.