Agentic Commerce: What Merchants Must Do Before AI Agents Become the Buyers

Agentic commerce is the model where AI agents discover, compare, and buy on a shopper's behalf, reading structured product data rather than marketing pages. To be selectable, merchants must make their catalog machine-legible: complete attributes and identifiers, real-time pricing and inventory, unambiguous shipping and returns terms, and support for the emerging commerce protocols. It is data discipline, not a rebuild.

For twenty years, ecommerce optimized for a human scanning a page. That assumption is breaking. AI agents are beginning to do the shopping themselves, interpreting a buyer's goal, comparing options, and in some flows completing the purchase, and they do not read your hero image or your persuasive copy. They read structured data. Merchants who make their catalog legible to an agent will be surfaced and bought; those who do not will be skipped before a human ever sees them. Here is what agentic commerce actually is, what merchants have to do now, and why this shift is one of the clearest vertical AI opportunities of the decade.

Agentic commerce is a model in which an AI agent interprets a shopper's goal, then autonomously researches, compares, and acts, up to and including completing a transaction, without the person manually browsing and clicking. Instead of a buyer opening ten tabs, the buyer tells an assistant what they want and the agent returns a short list or a completed cart. The shopper never sees the pages a merchant optimized. They see the answer the agent assembled from data it could read.

That is the core shift in how merchants must think about selling. The persuasion layer built for humans, the imagery, the copy, the page design, is largely invisible to an agent. What the agent consumes is structured product data: identifiers, attributes, price, availability, and fulfillment terms. A merchant whose catalog is precise, complete, and machine-readable becomes a candidate the agent can confidently select. A merchant whose data is thin or inconsistent is filtered out before consideration. This is the commerce version of a shift that has already reshaped content, and it is why commerce is one of the three sectors gAI Ventures co-founds companies in.

The infrastructure moved from theory to standard in a single year, and merchants have to know the names. The Agentic Commerce Protocol, developed by Stripe and OpenAI and open-sourced, lets agents read merchant product feeds and act on them, and it powers product discovery and merchant redirect inside ChatGPT shopping. Google and Shopify introduced a Universal Commerce Protocol aimed at letting agents interact with merchant catalogs, retrieve real-time pricing and inventory, manage carts, and complete purchases across platforms through one open standard, as covered in analyses of the AI trends shaping agentic commerce.

The takeaway is not to bet on one protocol. It is that the direction is now clear: agents expect a clean, machine-readable representation of your catalog and the ability to transact against live data. Whichever standards win, the underlying requirement, structured, accurate, current product data exposed in a way agents can consume, is the same, and it is the work worth starting now.

The good news is that this is data discipline, not a platform rebuild. The work concentrates in a few areas, and each one maps to a reason an agent includes or skips an offer.

Start with the audit. Pull your product feed and confirm every required and recommended field is filled, identifiers are present and correct, and the attributes a buyer would constrain on live in structured fields rather than in prose. Then fix the live-data layer, because agents deprioritize stale pricing and unclear availability to protect the buyer's trust in the agent. Finally, remove ambiguity from delivery, shipping, and returns, since an agent comparing offers quickly will pass over anything it cannot cleanly evaluate. None of this requires replatforming. It requires treating your product data as the primary sales surface it has become.

Here is where the shift matters beyond individual merchants. Making commerce agent-ready is not one generic task; it is deeply category-specific. What attributes matter, what a return signals, how pricing should flex, and what a product description must contain differ enormously between apparel, auto parts, grocery, and industrial supply. A horizontal tool that treats every product the same produces the generic output a category expert immediately distrusts. That gap, between what agents now require and what generic software provides, is exactly the terrain where vertical AI companies win, laid out for this sector in where vertical AI wins in commerce.

The scale makes the opportunity serious. McKinsey has estimated the global agentic commerce opportunity could reach several trillion dollars by 2030, and Shopify has reported orders from AI-powered searches growing many times over year on year. When a channel grows that fast and rewards category-specific data quality, the companies that own an agent-readiness workflow in a specific commerce vertical are building something a horizontal shopping assistant cannot easily copy. That is the kind of company gAI co-founds with expert operators, and the logic of backing domain experts to build these vertical companies is set out in the vertical AI investment theses and the gAI Ventures manifesto. Commerce companies built on this thesis sit in the gAI Ventures portfolio.

Two cautions keep this practical. First, do not bet the business on a single agent or protocol while the standards are still settling; the durable investment is in clean, structured, current product data, which pays off no matter which standards win. Second, do not treat agentic commerce as a far-future problem. The protocols shipped, the largest platforms are integrating, and the traffic is already growing, so the merchants who get their data house in order now will be the default answer when an agent assembles a shortlist, while laggards will be invisible in exactly the flows that are growing fastest.

For an operator who knows a commerce category deeply, this is more than a compliance exercise. It is the opening for a company. The workflows agents now demand, structured catalog data, live pricing, fulfillment clarity, reputation signal, are category-specific, judgment-heavy, and data-rich, which is the profile of a defensible vertical AI business. More on how gAI approaches building these companies is on the gAI Ventures blog, and the operators and engineers who build alongside founders are on the gAI Ventures team page. This article is educational thought leadership, not investment advice or an offer of any kind.