Agentic Commerce: How AI Is Changing Product Discovery and Buying

Picture a customer asking an AI assistant: “Find me a laptop for design work, compare five options, check availability, and help me buy it.” The interesting part here is not the conversation with the assistant. It is what happens to the stores selling those laptops: will their products even appear in that comparison? And is their data accurate enough for a purchase decision to rest on?

That is the essence of Agentic Commerce: an AI agent acting on behalf of the customer through part of the buying journey — discovery, comparison, evaluation, and possibly completing the purchase. This article explains what actually changed during 2026, why it matters to your store, and how to prepare with practical steps and no hype.

What Is Agentic Commerce?

In the simplest terms: an AI agent is software that acts on behalf of the customer. The customer sets the goal — “I need a gift for an eight-year-old within this budget, delivered before the weekend” — and the agent searches, compares, and recommends, then completes the purchase after the customer's approval or according to their prior instructions.

The technical term “Agentic Commerce” describes this pattern of trade: commerce in which an AI agent takes part in the buying journey. The idea is not a new store or a new app — it is a new interface between customer and merchant, added alongside search, websites, and apps, not replacing them overnight.

The Buying Journey Is Changing

Compare the two pictures below. The first is what we have known for years; the second is the emerging direction:

The traditional buying journey vs. the AI-assisted journey
Traditional journey	AI-assisted journey
The customer types keywords into a search engine	The customer describes their need in natural language or with an image
The customer browses sites and compares options themselves	The agent gathers options and compares specs, prices, and availability
The customer reads reviews and shipping and return policies	The agent summarizes reviews and policies and highlights the differences that matter
The customer chooses and pays inside the store	The customer approves the recommendation, and pays through a simplified flow — sometimes without leaving the conversation

One important caveat before continuing: this is an emerging direction, not a description of how most consumers shop today. The majority of purchases still follow the traditional path. But the pace of change during 2026 — as you will see below — makes early preparation a rational decision, not panic and not denial.

What Changed Recently?

Here are the key developments, documented from their official sources and ordered chronologically, with why each one matters to a store owner.

January 2026: The Universal Commerce Protocol (UCP)

Google, together with Shopify and around twenty partners — including Amazon, Meta, Walmart, Target, Etsy, and Wayfair — launched an open standard called the Universal Commerce Protocol (UCP). The idea: a common language letting platforms, agents, and merchants transact together — from product discovery to carts, payments, and order management — without custom integrations for every platform.

Why it matters to you: UCP is the infrastructure that could let a customer buy directly inside AI experiences such as AI Mode in Google Search and the Gemini app, while the merchant stays the Merchant of Record — owning the customer relationship and fulfilling the order.

2025–2026: Securing Agent Payments (AP2)

The hardest problem in agent-driven buying is payment: how does an agent pay on your behalf, safely? To answer it, Google developed the Agent Payments Protocol (AP2) — announced in September 2025 and donated to the FIDO Alliance in April 2026 to become an open industry standard. AP2 relies on cryptographically signed mandates that prove user intent, plus a non-repudiable audit trail when disputes arise.

In parallel, Mastercard launched Agent Pay (April 2025) and Visa announced Intelligent Commerce (April 2025), while Mastercard and Google co-developed the Verifiable Intent framework for documenting buyer authorization. The practical message: payment systems themselves are now being built on the assumption that an agent is party to the transaction — which is what makes agent-completed purchases technically possible, not just an idea.

May–September 2026: Google Prepares Merchants

In May 2026, Google started showing AI Performance Insights inside Merchant Center — reports that show merchants how their products are discovered across AI Mode, AI Overviews, and the Gemini app, benchmarking their visibility against competing brands. Then, on September 16, 2026, Google announced a seasonal update package: a UCP integration hub inside Merchant Center supporting cart transfer to the merchant's site and checkout-flow testing (with analytics coming soon), plus a Business Agent beta inside YouTube ads that lets viewers ask product questions without leaving the video.

Why it matters to you: Google says it plainly — the accuracy and richness of your product data in Merchant Center determines whether your products surface in AI recommendations. Google also reports that merchants following core feed best practices see an average 5% conversion uplift the following month.

June–September 2026: Shopify Moves In

In June 2026, Shopify announced its Spring '26 Edition: the Catalog API turning product catalogs into structured, agent-queryable data, UCP tooling for developers, and Agentic Storefronts surfacing merchant products inside ChatGPT, Copilot, and Google chats. Shopify reports that AI-driven traffic to its stores grew 8x year over year in Q1 2026, with orders from these channels up nearly 13x — Shopify's figures in their context, not a promise of similar results for your store.

In late September 2026, press reports (including PYMNTS and Dow Jones) described a Shopify–Meta partnership enabling Meta's Muse assistant — launched September 8 — to check out via Shop Pay across Shopify stores, after relying on Stripe's Link at launch. Details may evolve, but the direction is fixed: commerce platforms are racing to sit underneath every agentic transaction, even when the customer never visits the store.

Why This Matters to Online Stores

Because an agent — like a customer — cannot buy what it cannot find, understand, or trust. Here is what it actually depends on:

Product data: accurate names, descriptions, and specifications are what discovery and comparison are built on.
Pricing: any mismatch between the shown price and the checkout price loses the sale — and possibly the agent's future trust in your store.
Inventory: recommending an out-of-stock product is worse than not surfacing at all.
Delivery and shipping information: timing, cost, and coverage zones feed directly into the agent's decision.
Product attributes: size, color, material, compatibility — the details buyers ask about in natural language.
Policies: returns, exchanges, warranties, terms — the agent summarizes them for the customer, so ambiguity here takes you out of contention.
Checkout: every extra step or technical glitch means an abandoned customer — human or agent.
Payments: trusted, varied payment gateways raise completion rates from any channel.
Customer trust: genuine reviews and clear business information tip the scales in comparisons.
Structured, accessible information: the agent doesn't “admire” your site design — it reads your data. What can't be read can't be recommended.

Your Website May Need to Become “AI-Readable”

This phrase gets repeated a lot, and it means something simpler than it sounds: a beautiful website is not enough if important product information is hard to understand or retrieve automatically. The agent is not impressed by animations; it cares about finding an accurate answer to the customer's question.

In practice, a “readable” site is one where these elements exist — which happen to be the fundamentals of any good store:

Clean, complete product data: clear titles, honest descriptions, precise specifications.
Structured information (structured data): markup helping systems understand price, availability, and ratings.
Clear product attributes: sizes, colors, and options phrased consistently across the whole catalog.
Accurate, current prices and stock that reflects reality.
Shipping and return information written clearly and easy to find.
Standard product identifiers (such as GTIN or MPN) where applicable — the common language between systems.
Consistent information across your site, Merchant Center feeds, and every other sales channel.

And an honest note: there is no single universal standard for “AI visibility optimization” today. Anyone selling you a magic solution under that name is selling an illusion. The standing rule is: accurate, organized, consistent, accessible data.

What Changes for SEO?

The short answer: traditional SEO is not dead, but its job has expanded. Discovery can now happen through search engines, AI answers, AI shopping experiences, recommendation systems, or agents — often through a mix of them in a single journey.

Businesses therefore need both: traditional search visibility, plus trustworthy, machine-readable product and business information. Neglect the first and you lose the present; neglect the second and you risk the future.

What Changes for E-commerce Teams?

A practical checklist — review it quarterly:

Product catalog quality: every product with complete, accurate titles, descriptions, and specs.
Structured product data on product pages.
Inventory accuracy with real-time updates.
Price consistency across site, feeds, and channels.
Merchant feeds current and error-free.
Checkout reliability: regularly test the buying flow end to end.
Payment infrastructure: trusted gateways with options your customers use.
Policies published clearly: shipping, returns, warranties, privacy.
Analytics and conversion tracking: know where your customers really come from — including AI-experience traffic.
Customer data protected and organized — used with consent and under applicable rules.
Security: no unnecessary permissions for any external system, with periodic access reviews.

What Businesses Should NOT Do

Don't rebuild your entire store just because agentic commerce is trending — fix the foundations first.
Don't buy an “AI commerce” solution before understanding the exact problem it solves for you.
Don't publish auto-generated product information without human review — errors here hit trust directly.
Don't neglect traditional SEO on the grounds that agents are the future — the present is still mostly traditional search.
Don't give any agent or system unlimited access to customer or payment data.
Don't assume AI-driven traffic means automatically profitable customers — measure quality before celebrating volume.

How to Prepare Today: 7 Practical Steps

Each step below pays off even if agentic commerce grows slowly — that was the selection criterion:

Clean your product data: start with best sellers — accurate titles, honest descriptions, complete specs, real photos. This is the highest return on effort.
Improve product pages: answer buyer questions clearly (size? compatibility? shipping? returns?) before they are asked.
Ensure structured product information: enable structured data on product pages and review its errors periodically.
Keep inventory and pricing accurate: tie displayed stock to reality, and review any price conflicts across channels.
Improve checkout: fewer steps, test on mobile, and watch the cart-abandonment rate.
Review analytics and conversion tracking: make sure you measure orders and revenue accurately, and start monitoring traffic arriving from AI sources.
Follow AI-driven discovery developments: review quarterly what's new in Merchant Center and your platform's tools — without chasing every wave.

What Does This Mean for Rozanah Clients?

If your store or website runs with Rozanah, you already own the foundation: an organized site, improvable data, and a team tracking developments. The natural next step is a readiness review of your store — product-data quality, feed accuracy, checkout smoothness — then folding improvements into the regular development plan, not an emergency project.

And for companies considering automation and AI as part of digital transformation, agentic commerce illustrates a general rule: the value of any AI starts with the quality of your data. Clean the data first, and everything built on top becomes easier and more accurate.

Conclusion

The near-term opportunity is not “build an AI agent.” It is: make your business, products, data, and buying process ready for a world where AI may become another interface between you and your customers.

Start with data, secure the fundamentals, and follow developments calmly. The stores doing that today will be the same ones appearing in tomorrow's recommendations — whatever interface the customer arrives from.

Is Your Store Ready for the Agentic Commerce Phase?

Request an e-commerce readiness review covering: product-data quality, feed accuracy, checkout experience, digital infrastructure, and the automation opportunities that fit your business.

Request a review of your store's readiness to learn where you stand today — and which step pays back most for your business specifically.

Sources

The facts above come from their documented official sources. The review steps and recommendations are Rozanah's analysis from working with stores and websites — not official statements from Google, Shopify, or anyone else.