How retail is moving from shopping carts to shopping agents
Forget beautiful and eye-catching website design, A/B tests, and run-of-the-mill cart optimization.
In the world of agentic shopping, it all takes a back seat. AI-powered shopping agents aren’t browsing your homepage or responding to on-brand banner ads. They’re scanning product feeds, parsing metadata, and completing purchases autonomously. This coming paradigm shift demands that brands pivot their focus from visual storytelling to well-structured, rich product data.
The way consumers shop online is beginning to undergo a massive transformation. This new era sees AI tools gaining "agency" to act autonomously on a user's behalf to find, compare, and purchase products and services. This shift isn't in the distant future; it's already beginning, as evidenced by the Perplexity/PayPal integration announced in May 2025, plus tools being built by Visa, Mastercard, with others to follow.
Arguably, the most profound difference with agentic shopping is the disappearance of the traditional shopping cart and the entire checkout process as we know it. AI agents can navigate or completely bypass complex, and frequently tricky, checkout pages and apply stored payment information, then complete a purchase on behalf of a user.
Identity verification, CAPTCHAs, and the entire industry built around cart abandonment and retargeting ads also lose their relevance, as agents are not swayed or bypass these steps entirely.
Another major shift is the need to pivot from "traditional" Search Engine Optimization (SEO) toward what might be called Agent Feed Optimization (AFO) - ensuring product-level data is pristine, structured, and machine-readable.
Robust, complete data for every single product in a retailer’s catalog will power discoverability and personalization, giving agents the information they need to effectively act on behalf of shoppers.
Agents flip the script on sales funnels
In this agent-driven world, the current online shopping model is inverted. Traditionally, a human at the top of a shopping funnel narrows down their many choices to a specific product. With agentic commerce, a shopping agent, given a clear goal (e.g., "blue Nikes, size 10, under $100"), can instantly scan hundreds of merchant sources at the bottom of the funnel to find exactly what is needed. Agents are driven by price, specifications, and alignment with their user, not by emotional ads or lifestyle imagery.
A foundational technology enabling this is the Model Context Protocol (MCP), launched by Anthropic in late 2024. MCP is defined by Anthropic as "an open-source standard for AI-tool integrations" that allows AI models to connect and share data from various disparate systems like inventory, pricing databases, and customer profiles. Major players like OpenAI, Google, Shopify, Visa, and Amazon have rapidly adopted MCP, recognizing its role in allowing AI agents to access real-time data needed to function.
This shift will significantly disrupt merchants across five key areas:
- Search and discovery: Traditional SEO, user experience design, and visual merchandising lose ground as agents rely on structured metadata and product feeds to surface results.
- Purchase funnel: Transactions are initiated and completed within AI interfaces, reducing opportunities for upsells, cross-sells, and branded experiences.
- Brand loyalty: Loyalty may migrate to the AI layer, with agents reordering items on behalf of users who may not even recall the merchant brand.
- Platform shifting: New gatekeepers like OpenAI, Anthropic, and Perplexity determine which products are shown, shifting ad spend toward performance-based models.
- Merchandising tactics: Retail selling becomes “machines talking to each other.” Product data – feeds, reviews, pricing, inventory – must be semantic-ready and optimized for AI consumption. In this world, product data literally becomes marketing.
Monetization in this agentic world is crucial and often overlooked in current announcements. Obviously, agents don’t have eyeballs or emotions, so marketing budgets are likely to shift toward performance-based models , such as affiliate commissions or revenue shares. Sam Altman, CEO of OpenAI, has even speculated on charging a 2% affiliate fee for purchases made through their AI interface. This shift positions merchant-funded loyalty and rewards programs (including cashback) well, as they are funded by affiliate commissions.
Mass adoption of agentic commerce may be years away, but preparation can’t wait. Success will hinge on balancing convenience, trust, and value while building the infrastructure that powers it. APIs, product feeds, governance, and loyalty strategies, are all important components that will create the AI shopping platform of the future and position retailers to thrive in the agentic era.
Tristan Barnum is chief marketing officer and head of AI innovation at Wildfire Systems.
