🤖 AI Summary
AI companies and retailers are rolling out “agentic” shopping features—Instant Checkout in ChatGPT, Google’s form-filling agents, OpenAI’s Walmart partnership and deals with PayPal/Shopify—but executives and retailers say these agents are far from ready to fully handle purchases. Current implementations often require heavy user input, are slow or limited to a subset of items, and trade off data and control: AI firms want to protect chat histories and user intimacy, while retailers demand real‑time pricing, inventory and customer context. Tests show mixed results (Opera’s agent took 45 seconds to add eggs versus ~15 seconds manually), Expedia’s ChatGPT integration surfaces live pricing but still requires manual bookings, and Amazon is both suing Perplexity over agent purchases and piloting its own Buy for Me feature that shields merchants from shoppers’ real emails.
The stalling points are technical, commercial and legal. Agents need robust APIs for live inventory, low-cost compute for order execution, secure payment flows and clear data-consent models; retailers want commissions, fraud protections and accurate personalization. Negotiations involve payment processors and middleware startups trying to broker compromise. The result: consumer interest is high (surveys and McKinsey’s $1T-by-2030 projection), but practical, trustworthy end-to-end agentic shopping — with realtime context, correct pricing/delivery estimates and merchant buy-in — remains a work in progress, making human oversight likely for the holidays.
Loading comments...
login to comment
loading comments...
no comments yet