Agentic Commerce

Agentic commerce describes the emerging model where AI shopping agents handle the full purchase journey for consumers. Instead of a person searching Google, clicking product pages, and checking out manually, an AI agent does it all: parsing product catalogs, comparing options across retailers, and completing transactions through protocols like ACP, UCP, and MCP.

The term covers AI shopping across all major platforms: ChatGPT Shopping (900 million weekly active users), Google AI Mode, Perplexity, Claude, and emerging vertical agents. Morgan Stanley projects the AI-powered shopping market will grow from $3.36 billion to $28.54 billion by 2033 at a 27% CAGR.

Unlike traditional ecommerce where brands compete for human attention through design and marketing, agentic commerce is fundamentally data-driven. AI agents evaluate products based on structured attributes, not visual appeal. Retailers with richer, more accurate product data consistently win agent recommendations.

The agentic commerce flow has four stages:

Discovery: A consumer asks an AI assistant a natural language query ("find me waterproof hiking boots under $200"). The agent parses the intent and searches across product catalogs, feeds, and indexed web data.

Evaluation: The agent compares products on structured attributes: material, weight, waterproof rating, price, reviews, availability, shipping speed. Products with 30+ structured data points consistently outperform those with only 5-8 basic attributes.

Recommendation: The agent presents a curated shortlist. In Google AI Mode, 93% of these searches end without the consumer clicking through to a retailer's website. The recommendation itself is the moment of influence.

Transaction: Depending on the protocol and platform, the agent either completes checkout natively (via UCP in Google AI Mode) or refers the consumer to the retailer's site (ChatGPT's current model after OpenAI pulled back from native checkout in March 2026).

Agentic commerce runs on standardized protocols that define how AI agents interact with merchant systems:

Agent Commerce Protocol (ACP) was built by OpenAI with Stripe for ChatGPT Shopping. It handles product discovery through to checkout with Stripe as the payment rail. OpenAI shifted from native checkout to a referral model in March 2026.

Universal Checkout Protocol (UCP) is Google's standard, built with Shopify. It powers Google AI Mode shopping and has the broadest merchant adoption. Recent expansions include Nexi (Europe's largest paytech) and Splitit for BNPL.

Model Context Protocol (MCP) originated at Anthropic and now operates under Linux Foundation governance. It's a general-purpose AI-tool protocol increasingly used for commerce applications.

Agentic commerce adoption is accelerating across consumer segments and platforms:

• Market size: $3.36 billion in 2024, projected to reach $28.54 billion by 2033 at 27% CAGR (Morgan Stanley, 2026)
• Order growth: Shopify reports 15x increase in AI-originated orders year-over-year (Shopify, 2026)
• Consumer adoption: 23% of Gen X and 35% of Gen Z have used AI agents for shopping tasks (Forrester, 2026)
• Zero-click rate: 93% of Google AI Mode product searches end without the consumer clicking through to a retailer website
• Platform scale: ChatGPT has 900 million weekly active users; Google AI Mode is integrated into Google Search
• Enterprise investment: 90% of retailers are increasing AI budgets in 2026, with commerce applications as a top priority

Winning in agentic commerce requires a fundamentally different optimization strategy than traditional SEO or paid search:

Enrich product data: AI agents evaluate products on structured attributes. The minimum target is 30+ attributes per product, including material, dimensions, care instructions, compatibility, certifications, and use-case tags. Most retailers currently provide 5-8.

Optimize product feeds: Google Merchant Center feeds power Google AI Mode. Use product_detail attributes for everything beyond standard fields. Keep pricing and inventory data real-time.

Implement rich structured data: Expanded JSON-LD with additionalProperty fields, complete aggregateRating, offers with shipping details, and MerchantReturnPolicy schema.

Add llms.txt: This emerging standard tells AI crawlers what your site offers and where to find key information. Similar to robots.txt but for LLM crawlers.

Monitor AI visibility: Search for your products on ChatGPT, Google AI Mode, and Perplexity weekly using natural language queries your customers would use.

The discipline emerging around these practices is called Agentic Commerce Optimization (ACO).