The Rise of Agentic Commerce: What It Means for Brands When AI Becomes the New Front Door to Shopping

Until recently, a shopper would engage with your brand online in a predictable way: search, category browse, product page, cart, checkout.

That journey is compressing, fast. AI adoption has reached roughly 800 million global users in just three years, dramatically outpacing the early adoption curve of the internet. At the same time, mobile commerce and AI engagement are no longer siloed: nearly two-thirds of purchases now occur on mobile devices, over one-third of adults report having used AI guidance while shopping, and trust in agent-led purchasing is high among consumers who engage with it.

More consumers are beginning their shopping inside AI experiences that feel less like search and more like a conversation, and more of those conversations are ending in transactions. In this emerging model, AI does not just answer questions. It curates options, recommends products, builds carts, and in some cases completes checkout behind the scenes.

Danny Silverman, EVP, Digital and eCommerce, and Jenny Baker, Sr. Digital Content Director, have been spending a lot of time with MPG clients making sense of what actually matters right now, and what needs to change in day-to-day execution so brands are not caught flat footed later. They keep coming back to one idea: When AI surfaces fewer options, visibility stops being about winning a page and starts being about being understood in a structured, machine-readable way.

Their advice is not to chase every platform update. It is to make sure your product truth is complete, consistent, and current across the retailers that matter most. If you have ever watched a product lose share because of a broken item setup, a missing attribute, or outdated content, you already know how unforgiving the digital shelf can be. Agentic commerce just brings that risk closer to the moment the shopper decides.

What agentic commerce changes

Agentic commerce dramatically shortens the distance between “I need something” and “It’s on the way.”

Instead of keyword brainstorming, shoppers describe real situations in natural language: “What’s a low-sugar electrolyte for workouts?” “What batteries fit this toy?” “What’s a detergent for sensitive skin?” The AI interprets context, narrows options, and often recommends a small set, or a single best answer. In some cases, it can execute the purchase without the shopper ever visiting a website.

This reflects a shift from keyword-based search optimization (SEO) to agent answerability (AEO). Traditional shopping required browsing, scrolling, comparing, and evaluating. Agent-led shopping collapses those steps into moments. Research that once took tens of minutes can now be resolved in under a minute, with AI synthesizing information and acting on the shopper’s behalf.

To enable this at scale, the industry is standardizing how AI agents connect to commerce systems. OpenAI and Stripe have introduced the Agentic Commerce Protocol (ACP) to support agent-driven transactions while allowing merchants to retain control of payments and customer relationships. Google has introduced the Universal Commerce Protocol (UCP), designed to connect AI agents to discovery, cart, checkout, and post-purchase workflows without requiring retailers to rebuild their commerce stacks.

What’s happening across major platforms and retailers

Agentic commerce is emerging simultaneously across platforms, retailers, and ecosystems. The easiest way to stay grounded is to keep a simple view of how the major players are approaching agentic shopping and what that could mean for brands.

OpenAI: Instant Checkout inside ChatGPT

ChatGPT is evolving into what MPG describes as a “meta-mall:” a cross-retailer commerce layer where agents discover, decide, and transact using shared infrastructure. With Instant Checkout, shoppers can move from exploration to purchase without leaving the conversation. OpenAI has positioned results as relevance-based rather than ad-driven, while sellers retain control over payments and fulfillment.

Everyday shopping is already in scope. For example, Instacart’s integration enables meal planning, local inventory checks, and checkout directly inside ChatGPT, signaling that routine replenishment and pantry missions are central to this model.

In this environment, visibility is less about winning a scroll and more about winning the agent’s recommendation logic. Availability accuracy, pricing integrity, and attribute completeness increasingly determine whether a brand is surfaced at all.

Google: Gemini, AI Mode, and Universal Commerce Protocol

Google is transforming Search and Gemini into an end-to-end shopping agent powered by its Shopping Graph. Universal Commerce Protocol provides the connective tissue, allowing AI-driven discovery while keeping transactions anchored to retailer systems.

While the interface may feel simpler for shoppers, the underlying mechanics are evolving. Value signals, offers, and sponsored placements are expected to reappear in agent-native formats, with measurement shifting from clicks to agent-attributed outcomes.

Amazon: The closed-loop execution leader

Amazon currently operates the most advanced closed-loop agentic commerce system. Rufus and Alexa+ enable persistent goals, automatic buying, and frictionless reordering inside Amazon’s ecosystem.

However, Amazon’s experiments with surfacing off-Amazon listings highlight a critical challenge: clean data. Incorrect images, inaccurate pricing, discontinued items, and disruptive offers illustrate how fragile agent decisions become when product data governance breaks down.

Walmart: Open discovery, Walmart-controlled checkout

Walmart is making its catalog discoverable within Gemini and OpenAI, while at the same time investing in its AI assistant, Sparky, to keep site visitors in the Walmart ecosystem. Sparky is designed to build AI-driven shopping habits on site, while remaining open to external AI platforms off site.

For brands, the structural point matters: discovery can start in a third-party AI interface, but the retailer still controls checkout, fulfillment, and the customer relationship. Retailers want the reach of AI-driven discovery without giving up ownership of conversion.

Treat item setup and updates like growth levers, not maintenance work. If availability or key attributes are wrong or stale, you don’t just lose conversion, you risk not being recommended at all. Walmart has also described Gemini automatically including Walmart and Sam’s Club items when relevant to a shopper’s query, tying conversational discovery directly to retail assortment.

Target: AI-led discovery within a closed loop

Target is taking a similar approach, participating in UCP/ACP-enabled discovery across Gemini and AI Mode while keeping transactions within its own environment. As with Walmart, AI influences what gets considered, while the retailer retains control of conversion and loyalty.

Stepping back, the pattern is consistent. AI is becoming a primary discovery interface. Checkout may occur inside an AI experience, inside a retailer environment, or invisibly in the background. In every case, incomplete or inconsistent data doesn’t just reduce conversion—it eliminates visibility.

Key takeaway: Agent readiness is the new growth lever

Agentic commerce is scaling rapidly, but the biggest constraint is not AI sophistication, it’s commerce readiness.

Brands with clean, consistent catalogs, inventory, pricing, and fulfillment data gain an invisible but decisive advantage in agent selection. Ranking in AI mirrors the early days of Amazon and Google: strong governance, accurate taxonomy mapping, and structured attributes determine visibility.

The difference is that agents surface fewer options. The stakes are higher.

To win in AI-driven discovery, Marketing, Sales, Supply Chain, and Operations must align around launches and updates. If product truth lags behind real-world changes, the agent simply chooses someone else.

MPG recently helped a luxury laundry brand launch new SKUs on Amazon by optimizing product data specifically for Amazon’s AI agents, including Rufus and Cosmo AI. By structuring attributes, claims, use cases, shopper personas, and taxonomy for AI interpretability, the brand achieved strong initial discovery and conversion at launch. Subsequent optimizations to existing items improved both traditional SEO performance and Generative Engine Optimization (GEO), demonstrating that the same data fundamentals drive visibility across both search paradigms.

Where brands should start now

The goal is readiness with a little controlled testing, not a full-scale reinvention. Get the fundamentals right first, because those are what determine whether you show up. Then move quickly as retailer and platform programs expand in your categories.

Audit product data for AI interpretability Start with a basic question: If a shopper describes a real situation in plain language, do you have the attributes to match it? Focus on clean pack and variant logic, complete decision attributes, consistent taxonomy mapping, and claims that are clear and supportable.
Tighten attribute completeness and governance across retailers In agen- enabled shopping, structured fields do more of the work than brand prose. Make sure values and units are standardized, ownership is clear, and updates have a real workflow behind them. The goal is alignment between your brand source of truth and retailer item setup, not a patchwork of one-off fixes.
Treat availability and update cadence as part of the brand experience As discovery connects more directly to checkout, disappointment costs more. Strengthen item setup accuracy, reduce out of stocks on priority SKUs, and keep price and promo details clean, especially for bundles and multipacks. Packaging data matters here too, because it can determine whether an item is even eligible for certain fulfillment options.
Choose the first missions to optimize Do not try to perfect everything at once. Start where speed and confidence matter most: Replenishment, routine personal care and wellness, meal planning and pantry building, and seasonal missions. Pick the handful of SKUs that win those missions and make them the cleanest, most complete versions of your product truth.
Plan for measurement to shift Conversational shopping will not always produce the same clickstream you are used to. Build tests that isolate what changed, such as attribute completeness or governance improvements, and watch what happens to conversion and share. At the same time, start updating attribution thinking for journeys that begin in AI and end in retailer checkout.
Align teams around launches and changes A launch is no longer done when creative is ready. If pack architecture changes, claims change, sizes change, or inventory ramps, your product truth has to update everywhere quickly and consistently. Otherwise, the risk is not just a messy digital shelf. The risk is not being recommended at all when a shopper asks the question you should have won.

Is your content Agentic Commerce-ready? Do you know how your brands show up across Rufus, Sparky, or ChatGPT? Reach out if you're interested in an Agentic Commerce Safari, where we will audit your items and/or brands and set up time to take you through your current agentic presence, safari-style! Contact Danny Silverman, EVP, Digital and eCommerce, and Jenny Baker, Sr. Digital Content Director, for more information.