Agentic commerce is what happens when AI agents start buying things on behalf of your customers. Instead of a shopper visiting your site, scrolling through product pages, comparing options, and clicking "buy now," an AI agent does all of that for them. The shopper tells the agent what they want. The agent goes out, finds it, and the purchase happens without the shopper ever seeing your checkout page.
This isn't a chatbot answering questions or a recommendation engine suggesting products. In agentic commerce, the agent reasons through the decision, weighs trade-offs like price against shipping speed, and can complete the transaction on its own. The shopper stays in control by setting a budget, naming brand preferences, or approving the final purchase. But the browsing, comparing, and buying work shifts from human to machine.
A January 2026 IBM Institute for Business Value study surveyed more than 18,000 consumers across 23 countries and found that 45% already use AI for part of their buying journey. That covers everything from researching products to interpreting reviews to hunting for deals. The next step is agents that don't just research but act on what they find.
For store owners, this changes a basic assumption about how products get found and sold. Your site was built for human shoppers who browse, read descriptions, look at photos, and click buttons. Agentic commerce introduces a second type of buyer that doesn't interact with your store the way a person does. It reads your product data, evaluates your prices against competitors, checks your shipping times, and decides whether to buy from you or someone else.
How Is Agentic Commerce Different from Traditional Ecommerce?
The core difference is who does the work. In traditional ecommerce, the shopper handles every step of the buying process. They search for a product, visit several stores, compare prices, read reviews, add items to a cart, and enter their payment information. In agentic commerce, an AI agent takes over most of those steps. The shopper sets a goal. The agent figures out how to accomplish it.
Where AI Agents Replace the Browsing-and-Checkout Flow
An agent collapses the entire shopping journey into a single request. A shopper might say, "Find me running shoes under $120 that work for trail running and can arrive by Thursday." The agent searches across multiple stores, filters by price and features, checks inventory and shipping timelines, and presents a shortlist. If the shopper approves, the agent completes the purchase without ever loading a product page in a browser.
Your site's navigation, category structure, and visual merchandising matter less to agent traffic. What matters more is whether your product data is structured, accurate, and accessible through machine-readable formats. The agent doesn't see your hero banner or your lifestyle photography. It reads your product attributes, pricing, availability, and return policies.
What Changes for the Buyer and the Store
Buyers give up control of the browsing experience in exchange for speed. They don't see every option. They see what the agent recommends based on their stated preferences and the agent's evaluation of available products. Some shoppers will prefer this for routine or low-stakes purchases. Others will still want to browse, particularly for high-consideration items like furniture or electronics.
For stores, the shift means a new type of visitor is arriving that doesn't behave like human traffic. Agent visitors don't click through pages, don't respond to pop-ups, and don't read marketing copy. They query product feeds and structured data. Stores that only built for human visitors will miss purchases from agent-driven shoppers entirely.
| Traditional Ecommerce | Agentic Commerce | |
|---|---|---|
| Discovery | Shopper searches Google or visits a store directly | Agent queries product data across multiple stores simultaneously |
| Comparison | Shopper opens tabs and reads reviews manually | Agent evaluates price, reviews, shipping, and availability in seconds |
| Decision | Shopper chooses based on what they've seen and read | Agent recommends based on stated preferences and trade-off analysis |
| Checkout | Shopper enters payment details and clicks buy | Agent completes the purchase through delegated authorization |
| Post-purchase | Shopper tracks orders and handles returns manually | Agent can monitor shipments and initiate returns on the shopper's behalf |
How Do AI Agents Make Purchasing Decisions?
AI shopping agents decide what to buy by combining three capabilities that older commerce AI didn't have. They remember user preferences across sessions, reason through trade-offs between competing options, and pull real-time data from external systems. Together, these let an agent move from "here are some results" to "here's what you should buy and why."
Memory and Personalization
Agents remember what a shopper has told them and what they've bought before. If a shopper bought size 10 trail-running shoes last month and mentioned a preference for neutral cushioning, the agent stores that context. The next time the shopper asks for shoes, the agent doesn't start from scratch. It already knows the size, the preferred style, and possibly the price range the shopper is comfortable with.
For store owners, this makes your product data structure a direct factor in whether agents match your products to returning shoppers. If your listings don't include clear size information, activity type, or compatibility details, the agent can't connect your products to what it already knows about the buyer.
Reasoning and Trade-Off Analysis
Agents don't just filter results. They weigh competing factors against each other. If a shopper says "find me a camping tent under $150 that ships by Friday," the agent has to balance price, shipping speed, product quality, and availability. It might find a tent at $140 that arrives Saturday, and a tent at $155 that arrives Thursday. The agent reasons about which trade-off best matches the shopper's priorities based on what it knows about them.
This reasoning is what separates agentic commerce from basic search filters. A search engine returns results ranked by relevance. An agent returns a recommendation based on multi-variable analysis of what best fits the shopper's stated and inferred needs.
Tool Access and Real-Time Data
Agents pull live data from external systems to make informed decisions. They don't rely on cached information or static product listings. Through APIs and protocols like Model Context Protocol (MCP), agents can query a store's current inventory, check real-time pricing, verify shipping windows, and read return policies before making a recommendation.
If your inventory data is stale, your pricing isn't accessible through an API, or your product attributes are incomplete, the agent either skips your store or makes a recommendation based on bad data. Both outcomes mean you lose the sale, and the shopper's agent is less likely to come back to your catalog next time.
What Does the Agentic Commerce Ecosystem Look Like?
A new layer of infrastructure is forming between your store and your customers, and AI agents sit right in the middle of it. The ecosystem involves three types of interactions, a growing list of protocols that connect them, and a projected market that McKinsey estimates could reach $3 trillion to $5 trillion globally by 2030. For store owners, understanding where you fit in this system is the difference between getting found by agents and getting skipped.
Agent-to-Site Interactions
In the most common model, an AI agent interacts directly with a store's systems. The agent queries your product catalog, reads your pricing and availability data, evaluates your shipping options, and completes a purchase if the product fits the shopper's criteria. Think of it as a very fast, very focused shopper who reads your structured data instead of your web pages.
For this to work, your store needs to expose product information in a way the agent can read. That means structured product data, clean APIs, and accurate real-time inventory. If the agent can't pull your current prices or confirm an item is in stock, it moves to the next store.
Agent-to-Agent Interactions
In more advanced scenarios, your customer's agent negotiates directly with your store's agent. The shopper's agent might ask your agent for a bundle discount, check loyalty program eligibility, or negotiate delivery terms. Both agents act within pre-set rules, so the negotiation happens in seconds without either side needing human involvement.
This model is still emerging, but it's the direction platforms like Shopify and major retailers are building toward. Stores that have their own merchant-side agents will be able to participate in these negotiations. Stores that don't will be limited to passive catalog queries.
Protocols and Standards Shaping Agentic Commerce
A handful of open standards are being built to make agent-to-store and agent-to-agent interactions possible at scale. These protocols define how agents access product data, complete payments, and communicate with each other. None of them are fully mature yet, but they're moving fast.
| Protocol | What It Does | Who Built It |
|---|---|---|
| Model Context Protocol (MCP) | Gives agents structured access to external data like inventory, pricing, and product details | Anthropic |
| Agent-to-Agent Protocol (A2A) | Lets agents from different platforms negotiate and coordinate with each other | |
| Agentic Commerce Protocol (ACP) | Allows purchases to happen inside AI chat interfaces without leaving the conversation | OpenAI and Stripe |
| Agent Payments Protocol (AP2) | Creates auditable, cryptographically signed payment mandates for agent transactions | |
| Universal Commerce Protocol (UCP) | Covers the full commerce journey from discovery to post-purchase across platforms | Google (co-developed with Shopify, Etsy, Wayfair, Target) |
You don't need to implement all of these today. But you should know they exist, because they'll determine how agents find and buy from your store over the next one to two years.
How Do Agentic Payments Work?
Agentic payments are transactions where an AI agent, not a human, completes the purchase on a shopper's behalf. The shopper doesn't enter a credit card number or click a "buy" button. Instead, the agent uses pre-authorized payment credentials to pay for items within rules the shopper has already approved.
Delegated Authorization and Tokenized Credentials
The key mechanic behind agentic payments is delegated authorization. A shopper grants an agent permission to spend within defined limits, such as a maximum price per item, a monthly budget, or restrictions to specific product categories. From there, the agent uses tokenized credentials (a secure stand-in for the shopper's real payment details) to complete purchases without ever seeing actual card numbers.
Mastercard launched its Agent Pay program in April 2025, introducing Agentic Tokens that build on existing tokenization for contactless and card-on-file payments. Visa followed the next day with Intelligent Commerce, partnering with OpenAI, Anthropic, Stripe, and others to let agents transact using AI-ready tokenized credentials. Google released its Agent Payments Protocol (AP2) in September 2025, creating cryptographically signed payment mandates that link intent, cart, and payment together with a full audit trail.
For store owners, the practical impact is straightforward. Your payment system needs to accept transactions where the "buyer" isn't a human clicking through checkout. If your checkout flow requires human steps like CAPTCHA, pop-up confirmations, or manual form entry, agent-initiated purchases will fail at your store. The sale goes to a competitor whose checkout can handle it.
Fraud Detection When the Buyer Is Not Human
Traditional fraud detection looks for suspicious human behavior. Agent traffic doesn't follow those patterns. An agent might make a purchase at an unusual hour, from an unfamiliar IP, or without the browsing history that fraud systems use to verify legitimate shoppers. If your fraud tools aren't updated to recognize agent traffic, they'll flag or block legitimate purchases.
Payment networks are building solutions for this. Mastercard's program registers and authenticates trusted agents so stores can distinguish a verified AI shopper from a malicious bot. Visa is developing similar verification tools. But stores also need to check their own fraud settings, because an overly aggressive fraud filter will reject agent purchases and lose revenue without the store owner ever knowing it happened.
What Does Agentic Commerce Mean for Product Discovery and SEO?
Product discovery is splitting into two channels, and most stores are only visible in one of them. The first channel is traditional search, where human shoppers type queries into Google and browse the results. The second is agent-driven discovery, where AI agents query structured product data to find items that match a shopper's request. Stores that only invest in the first channel will become invisible to a growing share of buyer activity.
From Search Engine Optimization to Agent Optimization
SEO isn't going away, but it's no longer enough on its own. When a human shopper searches Google for "waterproof hiking boots," your SEO work gets your product page into the results. The shopper clicks through, reads your description, looks at your photos, and decides whether to buy. That entire process depends on your content being visible and persuasive to a person.
When an AI agent handles the same request, the process looks different. The agent doesn't visit your page in a browser. Instead, it reads your product data through structured feeds, APIs, or protocols like MCP. Your meta descriptions and lifestyle images don't register. What the agent cares about is whether your product attributes (price, availability, shipping speed, specifications, ratings) are accurate, complete, and machine-readable.
This shift has a name in the SEO world. It's called generative engine optimization (GEO), and it refers to making your content and data interpretable by AI systems, not just human readers. For ecommerce stores, GEO starts with your product data. If your catalog doesn't include standardized attributes that an agent can parse, your products won't show up in agent-driven recommendations.
Structured Data and Machine-Readable Catalogs
The single most important thing a store can do for agent visibility is clean up its product data. AI agents make decisions based on structured attributes, not marketing copy. They need specific fields to evaluate your products against a shopper's criteria.
The following fields are what agents look for in your product data.
| Field | Why the Agent Needs It |
|---|---|
| Product name and brand | Matches the shopper's brand preferences or category request |
| Price (current, sale, MSRP) | Compares against budget limits and competitor pricing |
| Availability and stock status | Skips out-of-stock items immediately |
| Shipping options and delivery time | Filters by the shopper's delivery deadline |
| Product specifications (size, color, material, compatibility) | Matches against stated requirements |
| Aggregate ratings and review count | Uses as a trust signal when comparing similar products |
| Return policy | Factors into risk evaluation for the shopper |
| Product category and type | Maps the product to the right intent category |
If you're running a Shopify or WooCommerce store, most of this data lives in your product listings and feeds. The gap is usually completeness. Many stores have pricing and availability but miss specifications, shipping details, or structured review data. Those missing fields are the difference between an agent recommending your product and recommending a competitor's.
Schema markup (specifically Product schema in JSON-LD format) is the most common way to make this data readable by AI systems. If your store already uses Product schema for Google Shopping or rich results, you're partway there. The next step is making sure every field is populated with accurate, current data and that your feeds update in real time so agents don't make recommendations based on stale inventory.
How Can Ecommerce Stores Prepare for Agentic Commerce?
The stores that show up in agent-driven purchases will be the ones that made their product data, systems, and checkout accessible to machines, not just humans. That work breaks into three areas, and none of them require a massive platform migration. Most of it builds on what you've already done for Google Shopping and structured data.
Make Your Product Data Machine-Readable
Start with your product catalog, because that's what agents read first. Every product listing needs complete, accurate, structured attributes. That means populating every field in your Product schema (JSON-LD format) with current data, including product name, brand, price, availability, shipping options, specifications, ratings, and return policy.
If you're on Shopify, WooCommerce, or BigCommerce, your platform generates some of this markup by default. The gap is usually in the fields it leaves empty. A listing with a price and a title but no availability status, no shipping estimate, and no review data gives the agent less to work with than a competitor who filled in everything. Completeness is what separates stores agents recommend from stores agents skip.
Open Your Systems with APIs
Agents need to pull live data from your store, and they can't do that if your systems are closed. At a minimum, your product catalog, pricing, and inventory should be accessible through an API that returns current data. If an agent queries your store and gets pricing from last week or inventory counts from yesterday, it might recommend a product you've already sold out of. That creates a bad experience for the shopper and a chargeback risk for you.
You don't need to build a custom API from scratch. Most ecommerce platforms have app ecosystems and integrations that expose catalog data. The goal is making sure the data the API returns is accurate and updates in real time.
Rethink SEO for AI Agents
Your existing SEO strategy gets your products in front of human searchers. Agent optimization gets them in front of AI buyers. The two aren't in conflict. They're complementary. But they require different inputs. Human-facing SEO depends on keywords, content quality, and backlinks. Agent-facing optimization depends on structured data completeness, feed accuracy, and protocol compatibility.
If you've been doing GEO work for AI search visibility, you're already partway there. The next step is checking whether your store's robots.txt file blocks AI crawlers like OAI-SearchBot, ClaudeBot, or PerplexityBot. Many ecommerce themes include default crawl rules that accidentally block these agents. A five-minute check can fix it, but most store owners never look.
What Are the Risks and Challenges of Agentic Commerce?
The biggest barrier to agentic commerce isn't the technology. It's trust. Shoppers are interested in using AI for purchasing decisions, but they're worried about what happens to their data and who controls their spending. A January 2026 IBM Institute for Business Value study found that 83% of consumers share concerns about privacy, data misuse, and unsolicited marketing from AI-driven commerce. That's a real gap between interest and comfort.
Consumer Trust and Data Privacy
Shoppers want the convenience of agent-driven purchases, but they aren't ready to hand over full control. Most consumers will start with low-risk tasks like reordering household items or comparing prices on products they've already researched. Fully autonomous purchasing (where the agent buys without the shopper approving each transaction) will take longer to gain adoption.
For store owners, trust affects whether shoppers authorize their agents to buy from you at all. If your store's privacy practices aren't transparent, or if your data handling policies are unclear, agents acting on behalf of privacy-conscious shoppers may avoid your store. Clear return policies, transparent pricing, and published data practices become not just good business but factors that influence whether agent traffic reaches your checkout.
Accountability When an Agent Makes a Bad Purchase
If an AI agent buys the wrong product, who handles the return? This question doesn't have a clean answer yet. If a shopper asks an agent for a medium blue shirt and the agent buys a large green one, the store gets a return. But the error wasn't the shopper's or the store's. It was the agent's interpretation of the request.
Payment networks and platforms are building frameworks for this. Mastercard's Agent Pay program includes processes for clarifying agent-initiated transactions. Google's AP2 creates audit trails linking the shopper's intent to the agent's action to the merchant's fulfillment. But for now, stores should expect to handle returns from agent purchases the same way they handle any other return, while keeping records of the transaction details in case dispute resolution processes evolve.
Where Is Agentic Commerce Happening Right Now?
Agentic commerce is live on several major platforms, and adoption is growing faster than most store owners realize. This isn't a concept that's five years away. Products are being discovered, evaluated, and purchased through AI agents today, and the infrastructure supporting these transactions launched over the past twelve months.
OpenAI introduced Operator in January 2025, an agent within ChatGPT that can browse websites, compare products, book reservations, and make purchases on a user's behalf. ChatGPT itself now has more than 900 million weekly active users as of early 2026, and OpenAI has integrated instant checkout through its Agentic Commerce Protocol (ACP), built with Stripe. Retailers including Etsy, Coach, Revolve, and more than a million Shopify merchants are live on ChatGPT's checkout integration.
Perplexity launched its "Buy with Pro" shopping feature in late 2024, allowing users to purchase products directly from search results. Google released the Universal Commerce Protocol (UCP) at the National Retail Federation conference in January 2026, co-developed with Shopify, Etsy, Wayfair, and Target, and endorsed by more than 20 partners including Visa, Mastercard, and Stripe.
On the payments side, Mastercard launched Agent Pay in April 2025, Visa introduced Intelligent Commerce the following day, and Google released AP2 in September 2025. All three are building the rails that let agents move money on a shopper's behalf.
For store owners, the question isn't whether agentic commerce will affect your business. It's whether your store is visible and accessible to the agents that are already shopping on behalf of your potential customers. The platforms are live. Shoppers are using them. And the infrastructure is here.
What Are the Most Common Agentic Commerce Mistakes Stores Will Make?
Blocking AI agent crawlers without knowing it. Many ecommerce themes and default robots.txt configurations block AI crawlers like OAI-SearchBot, ClaudeBot, GPTBot, and PerplexityBot. If these agents can't crawl your site, they can't index your product data, and they won't recommend your products. Check your robots.txt file and your server-side crawl rules. It takes five minutes and it's the single fastest fix you can make.
Treating structured data as optional. Plenty of stores have basic Product schema on their pages but leave critical fields empty. A listing with a title and price but no availability, no shipping estimate, no aggregate rating, and no return policy gives an agent almost nothing to work with. Agents compare products based on structured attributes. Incomplete data means your product loses to a competitor whose listing is fully populated, even if your product is better.
Assuming your current checkout works for agents. If your checkout requires pop-ups, CAPTCHA, multi-step form entry, or human verification steps, agent-initiated purchases will fail at your store. The agent can't fill out a CAPTCHA. It can't click through a confirmation modal. Stores that don't test their checkout for agent compatibility will lose sales without knowing it, because the agent simply moves to a store where the purchase goes through.
Ignoring agent-initiated payment support. Tokenized and delegated payment methods are how agents pay. If your payment processor doesn't support these transaction types, agent purchases can't complete. Talk to your payment provider about their roadmap for agentic payments and agent-initiated transactions. This isn't a feature you need to build yourself, but it is something you need to confirm your provider can handle.
Waiting for the technology to mature before doing anything. The platforms are live. Agents are already shopping. Payment rails are being built. Stores that wait for a "clear winner" among protocols or a "stable" ecosystem will miss the early adoption window. The preparation work (structured data, API access, crawl rules, payment compatibility) is the same regardless of which protocol wins. Start with the fundamentals and adjust as standards solidify.
Frequently Asked Questions About Agentic Commerce
What is the agentic commerce protocol?
The agentic commerce protocol (ACP) is a standard developed by OpenAI and Stripe that lets AI agents complete purchases inside chat interfaces without sending the shopper to a separate checkout page. When a shopper approves a purchase recommendation from ChatGPT, ACP handles the transaction within the conversation. It's one of several emerging protocols alongside Google's Universal Commerce Protocol (UCP) and Agent Payments Protocol (AP2). For store owners, ACP matters because it defines how your products get purchased through one of the largest AI platforms in the world. Stores connected through Stripe and Shopify integrations can already accept ACP transactions.
How is agentic commerce different from AI chatbots?
Chatbots respond to questions. Agents take action. A chatbot on your ecommerce site might answer "Do you have this in size 10?" but it can't go find the product, compare it to alternatives, and buy it. An AI agent can. Agents reason through multi-step tasks, access real-time data from external systems, and complete purchases using delegated payment credentials. The distinction matters because stores that built chatbot support haven't necessarily built agent support. Chatbots work within your site. Agents work across multiple stores simultaneously, making purchasing decisions based on structured data rather than on-page conversations.
Can small ecommerce stores benefit from agentic commerce?
Yes, small stores can benefit if their product data is structured and accessible. AI agents don't favor large retailers by default. They favor stores with complete, accurate, machine-readable product information. A small store with well-populated Product schema, real-time inventory data, and clear shipping and return policies can show up in agent recommendations alongside much larger competitors. The barrier isn't store size. It's data quality. Small stores that invest in structured data and keep their product feeds current will be discoverable by agents on the same terms as enterprise retailers.
What platforms support agentic commerce today?
Several major platforms are live with agentic commerce features as of early 2026. ChatGPT supports instant checkout through its ACP integration with Stripe, with more than a million Shopify merchants connected. Perplexity offers buy-ready search results through its "Buy with Pro" feature. Google launched the Universal Commerce Protocol at NRF 2026 with support from Shopify, Etsy, Wayfair, Target, and more than 20 payment and retail partners. Amazon's AI assistant Rufus handles product discovery for hundreds of millions of users. The landscape is moving fast, and new integrations are launching regularly.
How do AI agents choose which products to recommend?
Agents evaluate products based on structured attributes matched against the shopper's stated criteria. If a shopper asks for a waterproof jacket under $200 that ships in two days, the agent filters by waterproof rating, price, and delivery timeline. It then weighs secondary factors like aggregate review scores, brand reputation, return policy, and product specifications. Agents don't browse product photos or read marketing headlines. They read data fields. Stores with incomplete attributes (missing ratings, no shipping estimate, vague specifications) will rank lower in agent evaluations than stores with fully populated product data.
What is Model Context Protocol and why does it matter for stores?
Model Context Protocol (MCP) is a standard built by Anthropic that gives AI agents structured access to external data sources like product catalogs, pricing, and inventory. Instead of scraping your website, an agent using MCP can query your systems directly through a defined interface. For stores, MCP matters because it's the layer that connects your product data to AI agents in a clean, standardized way. If your data is accessible through MCP-compatible APIs, agents can pull accurate, real-time information about your products instead of relying on cached or incomplete web data.
Will agentic commerce replace traditional online shopping?
No, agentic commerce won't replace traditional shopping. It adds a second channel alongside it. Many shoppers will continue browsing stores directly, especially for high-consideration purchases like furniture, electronics, or clothing where they want to see the product and read detailed reviews. Agentic commerce will handle the purchases where speed and convenience matter more than the browsing experience, such as routine reorders, price-sensitive commodity products, and time-constrained shopping. Stores should prepare for both types of traffic rather than assuming one will replace the other.
How do agentic payments differ from regular online payments?
In agentic payments, the AI agent initiates and completes the transaction using delegated credentials instead of a human entering card details at checkout. The shopper pre-authorizes spending limits, and the agent uses tokenized payment methods (secure stand-ins for real card numbers) to pay. The key differences for stores are that there's no human clicking through checkout, no manual form entry, and no CAPTCHA. Your payment processor needs to support agent-initiated transactions, and your checkout flow needs to complete without requiring human interaction steps.
What structured data do AI agents need from product pages?
Agents need complete Product schema markup covering every attribute a shopper might care about. At minimum, that includes product name, brand, price (current and sale), availability, shipping options with delivery estimates, and product specifications. It also needs aggregate ratings with review count, return policy details, and product category. Most ecommerce platforms generate some of this markup by default, but the critical fields (shipping estimates, specifications, return policy) are often left empty. Fill in every field. Missing data is the most common reason an agent skips your product.
Is agentic commerce safe for consumers?
Yes, agentic commerce can be safe with the right guardrails in place, but the safety infrastructure is still being built. Payment networks like Mastercard and Visa are rolling out agent authentication systems that verify an AI agent is authorized to act on a shopper's behalf. Protocols like AP2 create cryptographic audit trails linking the shopper's intent to the agent's action. Shoppers retain control through spending limits, category restrictions, and approval requirements. The risk isn't the technology itself. It's the gap between what agents can do and the frameworks governing how they do it. That gap is closing, but it's not closed yet.
How does agentic commerce affect advertising and retail media?
Agentic commerce could reduce the effectiveness of traditional display advertising and retail media because agents don't see ads. When a human shopper browses your site or a marketplace, they see banner ads, sponsored listings, and promotional placements. An AI agent bypasses all of that. It reads product data, not marketing materials. McKinsey's research flags this as a significant risk for businesses that depend heavily on ad-driven revenue, particularly retail media networks. New monetization models will likely emerge, including sponsored placements within agent recommendation logic, but those models are still being defined.
What is agent-to-agent commerce?
Agent-to-agent commerce is when a shopper's AI agent negotiates directly with a store's AI agent to complete a transaction. Instead of the shopper's agent simply querying a product catalog, both agents communicate using protocols like A2A (Agent-to-Agent Protocol). The shopper's agent might ask the store's agent for a bundle discount, check loyalty program eligibility, or negotiate delivery terms. Both sides operate within pre-set rules, so the negotiation happens in seconds. This model is still emerging, but it represents the next stage beyond simple agent-to-site catalog queries.
How should stores handle returns when an agent made the purchase?
Handle agent-initiated returns the same way you handle any other return for now, but keep detailed transaction records. If an agent bought the wrong item on a shopper's behalf, the shopper still expects a smooth return process. The dispute isn't between you and the shopper. It's between the shopper and their agent. Payment networks are building resolution frameworks for agent-initiated transactions, but those frameworks aren't fully standardized yet. In the meantime, clear return policies published in structured data help agents make better purchasing decisions in the first place, which reduces return volume.
What role does brand play when AI agents make buying decisions?
Brand still matters, but it works differently when an agent is the buyer. A human shopper might choose Nike because they recognize the logo and trust the name. An AI agent evaluates brand through data signals like aggregate review scores, return rates, shipping reliability, and whether the brand's product data is complete and accurate. Strong brands with clean data, high ratings, and fast fulfillment will still win agent recommendations. But a lesser-known brand with better data quality and faster shipping can compete in ways that weren't possible when brand recognition alone drove the click.
How fast is agentic commerce adoption growing?
Adoption is accelerating faster than most projections anticipated. ChatGPT surpassed 900 million weekly active users by February 2026, and shopping is one of its fastest-growing use cases. McKinsey projects the global agentic commerce market could reach $3 trillion to $5 trillion by 2030. On the infrastructure side, every major payment network (Visa, Mastercard, Google, Stripe) launched agentic payment products between April and September 2025. More than a million Shopify merchants are already connected to ChatGPT's checkout system. The pace of change suggests that stores waiting for the market to "settle" before acting will find themselves behind.