How Retailers Should Prepare for Agentic Commerce

A shopper opens an AI assistant, types what they need, and an agent does the rest. It searches, compares, picks a product, and checks out. The shopper never visits a homepage, never sees your storefront, never reads your reviews. This is agentic commerce, and it is already live.

Agentic commerce is the model where an AI agent, not a human clicking through a website, does the searching, comparing, and selecting on a shopper's behalf. It moves the competition from winning clicks to being the product an agent picks. Different game, different rules.

We build and run product feeds for a living, so here is the part that matters: preparing for this is not a redesign and not a new channel. It is a data job. You need a catalogue an AI agent can read, trust, and buy from.

This is not a forecast. In 2026 the rails went live. OpenAI and Stripe co-developed the Agentic Commerce Protocol, an open standard that gives AI agents and merchants a shared language to complete purchases. Months later, Google launched the Universal Commerce Protocol, a competing open standard that covers the entire shopping journey and pulled major retailers and payment networks in within months.

The retailers are already live. U.S. shoppers can buy from Etsy and Wayfair directly inside Google's AI surfaces, with Shopify, Target, and Walmart next. On the OpenAI side, Etsy sellers went live first, with over one million Shopify merchants in the pipeline. One major European grocer even wired its full catalogue into an AI assistant, letting shoppers build a basket in chat. Shopify's own engineering team describes the protocol as the standard "enabling AI agents to connect and transact with any merchant."

When two competing protocols, the largest checkout companies, and the biggest retailers all standardise how machines buy in the same year, the discovery layer of e-commerce is being rewired. Retailers who prepared their data early become the default options agents reach for. Everyone else is invisible to a buyer that never visits a homepage.

When a person shops, your storefront persuades: imagery, layout, reviews, brand. When an agent shops, almost none of that reaches the decision. The agent reads structured data and decides in milliseconds, which changes what you optimise for.

• Your feed is your storefront. The agent evaluates titles, attributes, price, availability, and specs, so a beautiful page with a thin feed loses to an average page with a complete one.
• Ambiguity is rejection. An agent that cannot match an attribute to its filter skips the product, where a human would have guessed.
• Freshness is non-negotiable. A stale feed means oversells, cancellations, and an agent that distrusts you.
• Structured beats persuasive. Schema, clean taxonomy, and machine-readable specs win. Marketing adjectives can confuse matching.

Agentic commerce rewards operators over marketers, and we break down the buyer-side mechanics in how AI agents buy for customers.

Preparing for agentic commerce is a feed-engineering job, not a redesign. Here is the order we work.

1. 1.Complete every attribute that maps to a filter. Size, color, material, dietary flags, dimensions, gender, age group. An empty field is a missing sale.
2. 2.Standardise your taxonomy. Map categories to Google Product Category and a consistent internal tree. Inconsistent categorisation is the biggest reason agents fail to surface a product.
3. 3.Write titles for machines and humans. Lead with brand, product type, and key attributes in a predictable pattern. Drop promotional language.
4. 4.Sync price and stock in near real time so the agent never quotes a product you cannot ship.
5. 5.Add Product and Offer schema as a second, verifiable source of truth.
6. 6.Adopt the protocols. Where your platform supports the Agentic Commerce Protocol or the Universal Commerce Protocol, turn it on. One clean integration beats a dozen brittle connections.

For a full operational pass, we maintain a 27-step checklist to validate a product catalogue for agentic commerce.

The window to be early is short, and the moves are concrete. First, audit the feed you have. In the feed audits we run, it is common to find a third or more of a catalogue's attributes missing, inconsistent, or mapped to the wrong taxonomy, and every one of those gaps is a product an agent cannot confidently pick. Second, instrument freshness so price and stock update automatically, not in a nightly batch. Third, pick one agentic channel that fits your category and go live, whether a Shopify-native integration, a marketplace feed, or a direct partnership. Fourth, measure agent-driven traffic separately, because blending it into normal analytics hides the trend.

Retailers who do this in 2026 become the named options inside an agent's recommendation. The rest stay invisible.

Start with the data, not the channel. Every agentic surface reads from the same kind of source, so a complete, fresh, well-structured product feed is the one asset that prepares you for all of them. If you do one thing this quarter, run a feed audit and close the gaps.

There is also a compounding effect worth naming: the same clean feed powers your Google Shopping listings, your Meta catalog, your on-site search, and your structured data today. Nothing you fix for agents is wasted if agentic volume ramps slower in your category. That asymmetry is why feed work is the rare preparation with no downside scenario.

This is exactly what we build and run at L'Atelier Growth. We treat the product feed as critical infrastructure: complete attributes, standardised taxonomy mapped to Google Product Category, real-time price and stock sync, Product and Offer schema, and protocol-ready data an AI agent can read, trust, and buy from.

We are operators, not advisors. We get into your catalogue, fix the data, wire the syncs, and keep the feed healthy as inventory changes, so your store is a default option when agents buy. For a catalogue built for the agentic shift, start with product feed optimization.