You tell ChatGPT “find me a carry-on bag under $200 that fits United’s size limits, hard shell, arrives by Friday.” It asks two clarifying questions, searches five retailers, compares specs and reviews, and puts the best option in your cart. You approve with one click. That is agentic commerce, and it stopped being hypothetical in January 2026 when Google, Stripe, and Shopify launched the infrastructure to make it work at scale.
According to an IBM-NRF study from January 2026, 45% of consumers already use AI for at least part of their buying journey. Not just chatbot support or recommendation widgets. Actual product research, price comparison, and purchase decisions driven by AI agents that act on your behalf.
What Agentic Commerce Actually Means
Agentic commerce is when AI agents autonomously research, compare, negotiate, and complete purchases for users without constant human input. The difference from traditional e-commerce is fundamental. In traditional online shopping, you search, scroll, filter, read reviews, add to cart, enter payment details, and click “Place Order.” In agentic commerce, you describe what you want, and an agent handles everything from discovery to checkout.
The technical requirements are specific. An agentic shopping agent needs four capabilities that basic chatbots lack:
Memory: It remembers your preferences across sessions. You bought running shoes in size 11 last month; it does not ask again.
Reasoning: It evaluates options against criteria you set. Not just “cheapest” but “best value given my stated priorities.”
Tool use: It accesses real product catalogs, live inventory, current pricing, and shipping APIs. No hallucinated products.
Autonomous action: It can execute purchases within guardrails you define, like spending limits or approved merchants.
This is the shift from generative AI (which creates content) to agentic AI (which takes action). A recommendation engine suggests products. An agentic commerce system buys them.
Google’s Universal Commerce Protocol: The New Standard
On January 11, 2026, at the NRF Retail’s Big Show, Google CEO Sundar Pichai announced the Universal Commerce Protocol (UCP). It is an open standard that gives AI agents a common language to interact with merchant systems across the entire shopping journey: discovery, checkout, and post-purchase support.
Why UCP Matters
Before UCP, every AI agent needed custom integrations with every retailer. A shopping agent that worked with Amazon could not automatically work with Target or Zalando. UCP eliminates this by providing a single protocol that any agent can use with any participating merchant.
UCP is designed to work alongside existing agent protocols. It is compatible with MCP (Model Context Protocol) and A2A (Agent-to-Agent), the two foundational protocols for AI agent infrastructure. Where MCP connects agents to tools and A2A connects agents to each other, UCP specifically standardizes how agents interact with commerce systems: browsing product catalogs, checking inventory, processing payments, and handling returns.
Who Is On Board
The launch partners read like a retail industry directory. Google co-developed UCP with Shopify, Etsy, Wayfair, Target, and Walmart. More than 20 additional partners endorsed it at launch, including:
- Payment providers: Stripe, Visa, Mastercard, American Express, PayPal, Adyen
- Retailers: Best Buy, Macy’s, The Home Depot, Flipkart, Zalando
- Platforms: BigCommerce, commercetools, WooCommerce, Wix, Squarespace
Google immediately put UCP to work. The Gemini app and AI Mode in Search now offer checkout from participating U.S. retailers, powered by UCP. Target and Walmart customers can complete purchases without leaving Google’s AI interface.
The Platform War: Stripe, Shopify, and commercetools
Google built the protocol. But the platforms that power e-commerce backends are the ones making agentic commerce operational for merchants.
Stripe’s Agentic Commerce Suite
Stripe launched its Agentic Commerce Suite in January 2026, giving merchants a single integration point to connect their product catalogs to AI agents. The suite includes Shared Payment Tokens (SPTs) for secure agent-initiated transactions and Stripe Radar for fraud detection on non-human traffic.
Early adopters include Coach, Kate Spade, URBN (Anthropologie, Free People, Urban Outfitters), Revolve, and Ashley Furniture. Visa confirmed that hundreds of AI-initiated transactions have already been completed through the system, a sign that this is not vaporware.
Shopify’s Agentic Storefronts
Shopify co-developed UCP with Google and launched Agentic Storefronts, which merchants manage directly from Shopify Admin. The pitch to Shopify’s millions of merchants: enable AI agent access to your store without replatforming or rebuilding your catalog. The integration works with existing Shopify product data.
commercetools AgenticLift
commercetools, the composable commerce platform used by JD Sports and Nespresso, announced AgenticLift at NRF 2026. It is a standalone offering for agent-powered product discovery, cart building, and checkout, integrated with Stripe’s Agentic Commerce Suite. Enterprises can adopt it without replacing their existing commerce stack.
The AI Shopping Agent Battleground
On the consumer side, three companies are competing to become the default interface between shoppers and stores.
Google Gemini Shopping
Google has the most complete offering. Gemini can search product catalogs, compare options, and now complete checkout via UCP. It launched Business Agent, a virtual sales associate that brands like Lowe’s, Michael’s, Poshmark, and Reebok use to interact with customers through Google’s AI surfaces. Gemini Enterprise for Customer Experience extends this to The Home Depot, McDonald’s, Kroger, and Woolworths.
ChatGPT Shopping
OpenAI introduced shopping research as a guided buying experience in ChatGPT. It asks clarifying questions, pulls price and spec data, produces tailored buyer’s guides, and now offers Instant Checkout for participating merchants. Walmart was among the first to opt in. Shopping-related queries on ChatGPT rose 25% in just six months, from 7.8% to 9.8% of all prompts between January and June 2025, according to Bain & Company analysis. That translates to roughly 50 million daily shopping-intent queries.
Perplexity Shopping
Perplexity took a different approach. Its shopping experience uses conversational search with context retention, showing only details tied to your stated intent rather than infinite product grids. Perplexity partnered with PayPal for in-flow purchases, and merchants remain the merchant of record, handling returns and maintaining customer relationships directly.
The Numbers Behind the Shift
The market projections are substantial but grounded in observable trends:
- McKinsey estimates the agentic commerce opportunity at $3-5 trillion globally by 2030
- Morgan Stanley projects $190-385 billion in U.S. e-commerce spending by 2030, capturing 10-20% of the market
- 73% of consumers already use AI somewhere in their shopping journey, per eMarketer
- Only 12% of shoppers currently trust AI to make purchases autonomously, though 70% say they are at least somewhat comfortable with it
That trust gap between comfort and actual trust is the defining challenge. Consumers are willing to let AI agents research and recommend. Letting them buy requires a different level of confidence.
Security, Fraud, and the Trust Problem
Agentic commerce introduces fraud vectors that do not exist in human-driven shopping. Visa’s Payment Ecosystem Risk and Control (PERC) team reported a 450% increase in dark web posts mentioning “AI Agent” over a six-month period. The threats are specific:
Automated fraud at scale: Attackers can spin up thousands of AI agents to conduct targeted operations, testing stolen credentials or executing purchases across multiple merchants simultaneously.
AI-generated merchant fraud: Legitimate-looking websites with fabricated compliance documents and corporate identities, all created by AI agents.
Chargeback complexity: When an AI agent makes a purchase the customer disputes, who is liable? The customer, the AI platform, or the merchant? Existing chargeback processes assume a human buyer. Justt’s analysis flags this as an unresolved problem.
Prompt injection: Crafted inputs that bypass agent safety guardrails to extract payment data or trigger unauthorized purchases.
The industry response is emerging. Stripe’s Shared Payment Tokens limit what agents can charge. Visa and Mastercard are pushing Agent Commerce Protocol (ACP) and Agent Payments Protocol (AP2) for delegated authorization. The concept of “Know Your Agent” (KYA) verification, authenticating AI agents the way KYC authenticates human customers, is gaining traction as a security requirement.
What Happens Next
2026 is the proving year. The protocols are live. The platforms are integrated. The question is whether consumers will actually click “buy” inside AI interfaces.
The coexistence model is most likely. Agentic commerce will handle routine, low-stakes purchases: reordering household supplies, finding the cheapest flight that meets your criteria, replenishing consumables. High-stakes and emotional purchases, a wedding dress, a first car, a piece of art, will remain human-driven for the foreseeable future.
For retailers, the action item is clear: integrate with UCP and at least one agentic commerce platform (Stripe, Shopify, commercetools) before the 2026 holiday season. Merchants who are not visible to AI agents will be invisible to a growing share of shoppers.
For the DACH market specifically, Zalando has already adopted Google’s UCP, and OTTO’s Chief AI Officer is actively discussing AI agent integration at conferences like eTail Germany 2026. The EU AI Act’s full enforcement in August 2026 adds regulatory complexity: agentic commerce systems that make autonomous purchasing decisions may trigger high-risk AI classification, requiring transparency, human oversight, and GDPR compliance for personal data in agent transactions.
Frequently Asked Questions
What is agentic commerce?
Agentic commerce is when AI agents autonomously research, compare, and complete purchases on behalf of users. Unlike traditional e-commerce where shoppers manually search and buy, agentic commerce uses AI agents that act on instructions like “find me the best carry-on bag under $200 that arrives by Friday” and handle the entire process from discovery to checkout.
What is Google’s Universal Commerce Protocol (UCP)?
UCP is an open standard launched by Google on January 11, 2026, that gives AI agents a common language to interact with merchant systems. It enables agents to browse product catalogs, check inventory, process payments, and handle returns across any participating retailer. Launch partners include Shopify, Walmart, Target, Stripe, Visa, and Zalando.
How big is the agentic commerce market?
McKinsey estimates the global agentic commerce opportunity at $3-5 trillion by 2030. Morgan Stanley projects $190-385 billion in U.S. e-commerce spending through AI agents by 2030, capturing 10-20% of the market. As of early 2026, 45% of consumers already use AI for at least part of their buying journey.
Is agentic commerce safe?
Agentic commerce introduces new security challenges. Visa reported a 450% increase in dark web posts mentioning AI agents. Risks include automated fraud, prompt injection attacks, and chargeback disputes when AI agents make purchases. Mitigation measures include Stripe’s Shared Payment Tokens, Know Your Agent (KYA) verification, and Visa/Mastercard’s Agent Commerce Protocol.
Which AI shopping agents can I use today?
Three major AI shopping agents are available in 2026: Google Gemini Shopping (with checkout via UCP at participating retailers), ChatGPT Shopping (with Instant Checkout at merchants like Walmart), and Perplexity Shopping (with PayPal integration). Each takes a different approach, from Google’s full checkout to Perplexity’s guided conversational search.
We cover AI agents from protocols to production. Subscribe for practical guides every week.