Agentic commerce, explained for merchants

Key takeaways

• Agentic commerce means AI agents acting on delegated goals — shopping for buyers, operating for merchants

• Assistants like ChatGPT now surface products directly, compressing discovery into a short recommended list

• Machine-readable product data, accurate inventory, and fast pages decide whether agents recommend you

• Merchant-side agents make specialist growth work affordable, but human review before changes ship is non-negotiable

• The buyer side and merchant side are one shift: prepare your store for agents, and use agents to do the preparing

A working definition of agentic commerce

The simplest definition: commerce in which software agents, powered by large language models, are delegated a goal and then plan and execute the steps to reach it. That is different from automation you may already use. A rule that sends an abandoned-cart email is a fixed script; an agent decides what to do based on context, can chain multiple steps together, and can handle situations its author never explicitly programmed. In commerce, the delegated goal might be a shopper’s (“find me a waterproof jacket under $150 and order it”) or a merchant’s (“improve my product pages so they convert better”). The agent does the legwork; the human sets the goal and, in well-designed systems, approves the important decisions.

The buyer side: AI assistants that shop

The most visible half of agentic commerce is shoppers delegating research and buying to AI assistants. Instead of typing keywords into Google and opening ten tabs, a shopper describes what they need in a conversation, and the assistant compares products, surfaces a shortlist, and links — or in some flows, transacts — directly. The infrastructure arrived quickly: OpenAI and Stripe published the Agentic Commerce Protocol in late 2025, Google introduced its Universal Commerce Protocol with retail partners, and in March 2026 Shopify made products from eligible merchants discoverable inside ChatGPT by default. For merchants, the practical consequence is that an AI assistant is now a storefront surface you don’t control but can influence.

• Shoppers describe intent in natural language instead of composing keyword queries

• Assistants compare products across merchants and present a short, opinionated list

• Checkout increasingly hands off to the merchant’s own store rather than happening in-chat

• Product data quality largely determines whether your products get surfaced at all

The merchant side: AI agents that operate your store

The quieter half is merchants delegating operating work to agents. Running a store well means constant page building and editing, analytics review, SEO and AI-search visibility work, email and retention flows, and performance and security upkeep. Each of those is a specialty most small teams can’t staff. Agentic tools take a goal — grow organic traffic, lift conversion on key pages, reduce churn — then plan the tasks, execute them, and report back. The credible versions keep a human in the loop: the agent proposes concrete changes and the merchant reviews before anything ships. That review step is what separates delegation from losing control of your own store.

How agentic commerce differs from traditional e-commerce

Traditional e-commerce assumes a human at every step: a human searches, a human browses your page, a human decides, and on the merchant side a human plans and executes every change. Agentic commerce inserts software judgment into those loops. Discovery shifts from ranked links to an assistant’s synthesized recommendation, which compresses the consideration set — being one of three suggested products matters more than ranking eighth on a results page. Merchandising shifts from persuading human eyes to being legible to machines: structured data, accurate availability, and clear pricing become ranking factors in a very literal sense. And operations shift from doing tasks to directing and reviewing them.

• Discovery: from ten blue links to a short list chosen by an assistant

• Content: written for both human readers and machine readers at once

• Data: feeds, schema, and inventory accuracy become competitive assets

• Operations: merchants direct outcomes and review changes instead of executing every task

A short timeline of how we got here

The pieces landed fast. In 2025, ChatGPT added shopping-oriented product results, and in September 2025 OpenAI and Stripe released the Agentic Commerce Protocol as an open standard, initially powering an in-chat Instant Checkout pilot. Google followed with its Universal Commerce Protocol, developed with partners including Shopify, Target, and Walmart. By March 2026 the model shifted: OpenAI retired in-chat Instant Checkout and refocused ChatGPT on product discovery, while Shopify launched Agentic Storefronts, making roughly a million merchants’ catalogs discoverable in ChatGPT with checkout completing on the merchant’s own site. Meanwhile Shopify reported AI-driven traffic to its stores growing several-fold year over year through 2026.

What this means for small and mid-sized merchants

Two things are true at once. First, buyer-side agents reward preparation over ad budget: clean product data, honest pricing, fast pages, and content that answers real questions are what get you recommended, and those are achievable at any size. Second, merchant-side agents lower the cost of doing that preparation well — work that used to require an agency or several hires can now be delegated. That’s the model behind WorkDaddy: you set a growth goal, specialized agents for site building, analytics, visibility, retention, and operations plan and execute the work, and you review changes before they go live. The merchants who treat both sides as one shift, rather than two separate trends, will compound the advantage.