What investors missed at Shoptalk 2026: AI is starting to own the checkout

Something important shifted at Shoptalk 2026, and most investors might miss it at first glance. It wasn’t about foot traffic or whether shoppers feel good about spending.

Retailers are no longer treating AI like a side tool for product discovery. They’re building for it as a place where purchases actually happen. Gap is preparing for buying directly inside Google AI Mode and Gemini using Google’s Universal Commerce Protocol, while moves like Walmart and Google’s AI-powered shopping push show how quickly major retailers are adapting. Sephora already has an app inside ChatGPT with loyalty perks built in, and it’s moving toward full checkout and payments right there. At the same time, both Google and OpenAI are rolling out the infrastructure to support this shift, turning AI interfaces into full commerce channels.

It’s tempting to see this as just the next step in search. People used to type into Google, now they ask a chatbot. But that framing misses what’s really changing.

AI can now handle the entire shopping journey in one place. It can help you discover a product, recommend options, apply your loyalty rewards, surface promotions, and complete the transaction without ever sending you somewhere else. When that happens, AI stops being a traffic source and starts becoming the layer that controls distribution. In some markets, that model is already taking shape through AI apps that connect shopping, travel, and payments together.

If AI decides what gets seen, how products are ranked, which perks apply, and where checkout happens, it holds real power over demand. For retail investors, this isn’t just about better search. It’s about who owns the customer relationship and who gets pushed to the sidelines as AI becomes the new storefront. That competitive scramble is becoming more visible as Meta tests AI shopping tools to compete with OpenAI and Google.

Why machine-readable commerce could become retail’s new tollbooth

In the search era, a merchant could lose the first click and still keep the core structure of commerce.

The brand site held the catalog, the customer account, the basket, the promotion logic, and usually the checkout. In AI commerce, that stack can be pulled apart. The surface interpreting intent can also hold customer context, decide which products to show, call structured catalog feeds, apply loyalty, and route the user either into merchant-owned checkout or into a platform-mediated flow. That is why protocols matter. They are not just plumbing. They are the control plane for who gets to participate in the transaction and on what terms.

Gap’s move makes that plain. The company said outright that shopping is shifting from traditional search engines to AI-powered answer engines, and it is responding by making its products transaction-ready inside conversational environments. Through UCP, customers will be able to buy Gap products while using AI Mode in Google Search and the Gemini app. This is not a content optimization project. It is a distribution decision. Gap is choosing to make its catalog, sizing, and checkout legible to the machine now sitting between buyer intent and merchant conversion.

Google’s own product design makes the stakes even clearer. UCP is built to support direct buying on AI surfaces with real-time product details, live inventory, multi-item carts, and identity linking so loyalty benefits can follow the shopper. Google is also careful to say the merchant remains the merchant of record and keeps the customer relationship. That reassurance is telling. Platforms do not stress merchant control unless merchants are worried about losing it. The issue is whether the interface handling discovery also starts to intermediate identity, merchandising, and transaction flow.

Sephora is the strongest proof that retailers know this is not a one-platform bet. Its ChatGPT app is not framed as a novelty assistant. It already lets users discover and shop with guided advice while accessing Beauty Insider rewards and benefits, and Sephora has said direct payments and checkout are coming inside the app. That is what an AI-native retail presence looks like when a merchant takes the channel seriously: not just personalized recommendations, but loyalty continuity and eventual transaction execution inside the conversational layer itself. The same surface advising the shopper begins to shape conversion.

OpenAI’s current posture sharpens the point. The company is expanding structured shopping and agentic commerce tools, but it has also stepped back from pushing a fully standalone Instant Checkout model and instead prioritized merchant-owned checkout. That is not evidence that AI commerce is weak. It shows where the fight already is. Platforms know the discovery layer is valuable. Merchants know checkout and identity are more valuable still. And as ChatGPT’s ad business begins to scale, the economics of controlling that discovery layer may become even more attractive.

What to watch next

The first signal is whether more large merchants launch AI-native loyalty and checkout integrations instead of stopping at AI shopping assistants.

Gap and Sephora matter because both moves reach into commercially sensitive territory: product data, rewards, and the transaction itself. If more brands expose loyalty balances, promotions, baskets, and fulfillment options inside AI interfaces, that will suggest this is becoming operating infrastructure rather than conference theater.

The second signal is whether commerce protocols harden into standards that merchants treat as required distribution work. Google is simplifying UCP onboarding in Merchant Center and adding features such as multi-item carts and identity linking. OpenAI is expanding structured commerce tooling for product discovery and agentic flows. If merchants start talking about protocol implementation the way they once talked about mobile app launches, feed management, or marketplace integrations, that will mark a real shift in retail operating assumptions. The same dynamic is also playing out internationally, where Chinese tech giants are racing to build AI agents inside super apps.

The third signal is whether companies start reporting the channel as a channel. Watch earnings calls, investor commentary, and executive interviews for specific references to AI-referred traffic, conversion quality, basket size, merchandising changes, or owned-checkout performance from AI surfaces. Watch hiring and budget shifts too. When brands start moving resources from classic SEO and landing-page optimization toward structured product data, AI-surface relevance, and in-surface commerce operations, that will be the tell. Without those signals, this remains a promising emerging layer. With them, it starts to look like a new gatekeeper layer in retail.

Search is not dead. Brand sites are not disappearing. Retail does not need either to vanish for the economics to change. It only needs a new layer to sit in front of them and start influencing who gets seen, who gets the data, and who closes the sale. That layer is starting to form now, and the shape of that future is becoming clearer across Western platforms and Asian super-app ecosystems alike.