For twenty years, online shopping has followed the same path: a customer searches, clicks through to a store, browses, adds to cart and pays. In 2026 a new step is appearing at the front of that journey. Shoppers ask an AI assistant such as ChatGPT, Google's AI Mode or Microsoft Copilot to find "a waterproof hiking jacket under $150 that ships this week", and the assistant does the searching, comparing and shortlisting for them. In some cases it can complete the purchase too. This shift is called agentic commerce, and it changes how customers find your products.
The big platforms have moved quickly. Google launched an open standard for agent-led shopping in January 2026, Microsoft added checkout inside Copilot the same week, and OpenAI and Stripe published their own protocol in late 2025. Adobe's data shows AI-referred visits to US retail sites growing fast and converting better than other traffic. If an AI agent cannot read your catalogue, check your stock or trust your prices, it will recommend a competitor instead, and you may never know the sale was lost.
This guide explains what agentic commerce is, how AI shopping agents work, the protocols behind them, what actually happened in 2026, and a practical checklist and 90-day roadmap to get your store ready. It is written for store owners and e-commerce managers, with the technical parts explained in plain language.
Agentic Commerce at a Glance
Quick answer: Agentic commerce is shopping in which an AI agent acts for the customer: it understands what they want, searches and compares products across stores, answers questions and helps complete the purchase. To be chosen by these agents, your store needs complete, accurate, machine-readable product data, real-time stock and pricing, clear policies, and connections to the platforms and protocols the agents use.
- Standards: OpenAI and Stripe released the Agentic Commerce Protocol (ACP) on 29 September 2025; Google launched the Universal Commerce Protocol (UCP) on 11 January 2026.
- Platforms: ChatGPT, Google AI Mode and the Gemini app, and Microsoft Copilot all support AI-assisted shopping.
- Traffic: Adobe reported that AI-referral traffic to US retail sites rose 62% year over year in July 2026, and converted 60% better than non-AI traffic.
- Lesson so far: OpenAI moved its Instant Checkout into merchant apps in March 2026 and is prioritising product discovery. For most stores, the near-term win is being found and recommended by AI, then converting on your own site.
- First steps: fix your product data, expose accurate stock and prices, enable the channels your platform supports, and measure AI traffic.
What Is Agentic Commerce?
Agentic commerce is e-commerce in which software agents, rather than people, do much of the work of shopping. An "agent" is an AI system that can take actions toward a goal, not just answer questions. In shopping, that means an AI agent can interpret a request, search many sources, compare options against the shopper's needs, check availability and delivery, and in some setups place the order with the shopper's approval.
It helps to separate three related ideas:
- AI-assisted discovery: the assistant recommends products and links to the store, where the customer buys as usual. This is the most common form today.
- In-chat checkout: the customer confirms the purchase inside the assistant, and the merchant still processes the order and payment.
- Delegated buying: the customer gives the agent rules ("reorder coffee when I run low, under $20") and the agent buys within those limits. This is still early.
In every case, the merchant usually stays the seller of record: you still own the customer relationship, fulfilment, returns and support. What changes is who, or what, is browsing your store.
Why Agentic Commerce Matters in 2026
Agentic commerce matters in 2026 because the largest AI platforms and payment companies have built the infrastructure, and shoppers are already arriving through it. Three developments stand out.
1. Shared standards now exist
Agents need a common language to talk to thousands of stores. Two open standards emerged: OpenAI and Stripe's Agentic Commerce Protocol in September 2025, and Google's Universal Commerce Protocol in January 2026. Google says UCP was co-developed with Shopify, Etsy, Wayfair, Target and Walmart and endorsed by more than 20 companies, including Adyen, American Express, Best Buy, Flipkart, Macy's, Mastercard, Stripe, The Home Depot, Visa and Zalando. When retailers, platforms and card networks agree on standards, adoption tends to follow.
2. AI assistants are sending real shoppers
Adobe Analytics tracks visits to US retail websites. According to Digital Commerce 360's reports of Adobe's data, traffic from generative AI sources to US retail sites rose 693% year over year across the 2025 holiday season, and in July 2026 AI-referral traffic was up 62% year over year. Adobe also found that these AI-referred visits converted at a rate 60% higher than non-AI traffic and generated 53% more revenue per visit. AI is still a smaller channel than search or social for most stores, but it is growing quickly and bringing shoppers who arrive ready to buy.
3. Checkout is moving closer to the conversation
Microsoft launched Copilot Checkout on 8 January 2026, letting US shoppers buy from participating merchants inside Copilot, with PayPal, Shopify and Stripe as launch partners. Google's UCP powers checkout in AI Mode and the Gemini app for eligible US retailers. Even where the final purchase happens on your website, the decision is increasingly made inside the AI conversation.
What this means for you: the "shelf" where customers compare products is moving into AI assistants. Stores with clean data and connected systems get recommended; stores that agents cannot read get skipped.
How AI Shopping Agents Find and Buy Products
An AI shopping agent follows a sequence that mirrors what a careful human shopper does, only faster and across more stores:
- Understand intent. The agent turns a request such as "a gift for a runner who has everything, under $80" into requirements: category, budget, attributes and timing.
- Discover products. It searches product feeds shared with the platform, the open web and structured data on product pages.
- Compare and filter. It checks attributes, reviews, price, stock, shipping times and return policies, and drops anything it cannot verify.
- Recommend. It presents a short list with reasons, and answers follow-up questions about size, materials or compatibility.
- Transact. Depending on the platform, it links the shopper to your product page or checkout, or creates a checkout session through a protocol such as UCP or ACP, using a payment token so card details are not shared with the agent.
- Follow up. Order status, delivery updates and returns flow back through the platform or your own channels.
Notice where agents fail: missing attributes, out-of-date stock, unclear delivery times and vague return policies. Each gap is a reason for an agent to choose another store.
The Protocols: UCP, ACP and MCP Explained
Protocols are the agreed rules that let any agent talk to any store without a custom integration each time. You do not need to read the specifications, but you should know what each one does.
| Protocol | Who is behind it | What it covers |
|---|---|---|
| Universal Commerce Protocol (UCP) | Google, co-developed with Shopify, Etsy, Wayfair, Target and Walmart (January 2026) | The full journey from discovery and buying to post-purchase support; powers checkout in Google AI Mode and Gemini for eligible US retailers |
| Agentic Commerce Protocol (ACP) | OpenAI and Stripe (September 2025) | How agents and merchants exchange product, checkout and payment information; open standard usable with other payment providers |
| Model Context Protocol (MCP) | Open standard for connecting AI applications to tools and data | A general way for AI agents to call your systems, such as product search or order status; Google says UCP works with it |
Two details matter for merchants. First, both UCP and ACP are designed to keep the merchant as the seller of record, so you keep control of pricing, fulfilment and the customer relationship. Second, payments use scoped tokens: Stripe describes its Shared Payment Token as a way for an application like ChatGPT to start a payment without exposing the buyer's card details, limited to a specific merchant and amount.
If you run Shopify, much of this is handled by the platform's own channels. If you run a custom store, a headless build or a marketplace, you will need development work to expose your catalogue, inventory and checkout in a way agents can use. That is where software and API integration comes in.
Where AI Shopping Happens: ChatGPT, Google and Copilot
ChatGPT
OpenAI launched Instant Checkout in ChatGPT in September 2025, starting with US Etsy sellers and announcing support for Shopify merchants. In March 2026, an OpenAI spokesperson told Digital Commerce 360 that "Instant Checkout is moving to Apps, where purchases can happen more seamlessly", and that OpenAI is "prioritizing making ChatGPT search and product discovery great, with ACP serving as the infrastructure that connects users to merchants across the full shopping journey." In practice, ChatGPT is a major discovery surface, and merchants can share product information through ACP and build experiences through ChatGPT apps.
Google AI Mode and Gemini
Google's UCP powers a checkout feature in AI Mode in Search and the Gemini app, letting shoppers buy from eligible US retailers with Google Pay or PayPal while the retailer remains the seller of record. Google's Merchant Center product data remains central to how products appear.
Microsoft Copilot
Microsoft says Copilot Checkout "turns conversations into conversions" with no redirect, while the merchant stays the merchant of record. At launch, Shopify merchants were enrolled automatically after an opt-out window. Microsoft also introduced Brand Agents, which it describes as AI-powered shopping assistants that speak in a brand's voice.
The pattern across all three: your product data feeds the discovery, and your systems still fulfil the order.
The 2026 Lesson: Discover in AI, Buy on Your Site
The most useful lesson from the first year of agentic commerce is that discovery has moved faster than checkout. OpenAI's decision to move Instant Checkout into apps and focus on product discovery shows that fully automated buying is harder than it looks. Real stores have promotions, bundles, loyalty pricing, size variants, store pickup and stock that changes by the minute, and agents must handle all of that reliably before shoppers and merchants trust them with the whole purchase.
For most businesses, that leads to a clear priority order:
- Be discoverable. Make sure AI agents can find, read and trust your products.
- Convert the click. Make the landing experience fast and relevant for shoppers who arrive from an AI recommendation with a specific product in mind.
- Enable in-AI checkout where it is available and sensible. Turn on the platform channels that fit your products and margins, and test them.
This is good news for smaller retailers. You do not need to build a cutting-edge agent integration on day one. You need excellent product data and a store that converts.
Product Data Is Your New Storefront
When a human browses, good photos and persuasive copy carry the sale. When an AI agent browses, structured facts do. Agents compare attributes, and they cannot recommend what they cannot verify. Strong product data for agentic commerce includes:
- Specific titles that state what the product is, for example "Men's waterproof hiking jacket, recycled nylon, navy" rather than a brand-only name.
- Complete attributes: size, colour, material, dimensions, weight, compatibility, age range and certifications you can prove.
- Accurate price and stock, updated frequently, including variant-level availability.
- Shipping and returns stated clearly per market: delivery times, costs and return windows.
- Unique identifiers such as GTINs where they exist, so agents can match your products across sources.
- Structured data (schema.org Product, Offer and review markup) on product pages, plus feeds for Google Merchant Center and other platforms.
- Answers to common questions written into product pages: fit, care, what is in the box, warranty.
If your catalogue lives across an ERP, a PIM, spreadsheets and the store itself, the hardest part is often keeping one accurate version. Connecting those systems so data flows automatically is usually the highest-return technical investment in agentic commerce.
How to Get Your Store Ready for Agentic Commerce
Use this checklist to judge how ready your store is today. Each item improves AI visibility and usually your normal search and conversion too.
- Audit your product data. Find missing attributes, duplicate products, vague titles and stale prices. Fix the best-sellers first.
- Publish accurate feeds. Keep Google Merchant Center and other feeds complete and refreshed often enough to match real stock.
- Add structured data. Make sure product pages carry valid schema markup for products, offers, availability and reviews.
- Enable platform channels. If your platform offers agentic or AI channels, review the terms, enable the ones that fit, and decide which products to include.
- Expose real-time availability. Connect inventory and pricing so agents and your site never promise stock you do not have.
- Make policies machine-readable. Put shipping, returns and warranty terms on clear, crawlable pages and in your feeds.
- Optimise the landing experience. Shoppers from AI arrive with intent; send them straight to the right product with fast pages and simple checkout.
- Measure AI traffic. Segment visits referred by AI assistants in analytics, and track their conversion and revenue separately.
- Prepare your systems for agents. If you run a custom or headless store, plan APIs for catalogue, cart and order status that can support UCP, ACP or MCP-based integrations.
- Train your team. Support, merchandising and marketing teams should know how AI channels work and how to spot problems.
Build Your Own AI Shopping Assistant
Agentic commerce is not only about appearing in other companies' assistants. The same technology can work on your own website, app or WhatsApp channel. An AI shopping assistant grounded in your catalogue can answer product questions, recommend items based on a shopper's needs, check order status and hand complex cases to a person.
Good use cases include:
- Guided selling for products with many options, such as electronics, furniture, beauty or B2B parts.
- Pre-purchase questions about sizing, compatibility, delivery and returns, answered instantly at any hour.
- Order and returns support connected to your order management system.
- Conversational commerce on WhatsApp, where many customers already prefer to talk to brands. Our WhatsApp marketing guide covers this channel.
The key is grounding: the assistant must answer from your real product data, stock and policies, not guess. That requires integration with your catalogue and systems, plus testing and monitoring. CodeBase Coders builds these as AI chatbots and, where they take actions such as creating returns, as AI agents.
Risks, Trust and Compliance
Agentic commerce brings new risks that are worth planning for:
- Wrong information at scale. If your feed shows the wrong price or stock, an agent can repeat that mistake to many shoppers. Accurate, automated data is the main defence.
- Less control over presentation. Agents summarise your products in their own words. Clear, factual product content reduces the chance of misrepresentation.
- Fraud and authorisation. Payment tokens reduce exposure of card data, but you still need fraud checks and clear rules for agent-initiated orders.
- Platform dependence. AI channels can change quickly, as 2026 showed. Keep your own site, customer data and direct channels strong.
- Privacy and consumer law. Customer data shared with platforms and assistants is still subject to privacy rules where you sell, and consumer protection rules on pricing, returns and disclosures still apply.
What Agentic Commerce Readiness Costs
There is no single price, because the work depends on your platform, catalogue size and systems. The main cost factors are:
- Platform: a standard Shopify store can use built-in channels with little development, while a custom, headless or multi-system store needs integration work.
- Catalogue size and quality: a few hundred well-described products is a small clean-up; tens of thousands of SKUs with poor data is a larger data project.
- Number of systems: connecting an ERP, PIM, inventory and order management system adds integration effort.
- Custom AI features: an on-site shopping assistant or agent adds design, development, testing and running costs, including AI model usage.
- Ongoing work: feeds, integrations and AI features need monitoring and maintenance as platforms change.
A sensible approach is to start with a short readiness audit, fix product data and feeds, measure results, and only then invest in deeper integrations or custom agents.
A 90-Day Agentic Commerce Roadmap
Days 1 to 30: Assess and fix the basics
- Audit product data for your top products and categories.
- Fix titles, attributes, identifiers, prices and stock accuracy.
- Validate structured data and Merchant Center feeds.
- Set up analytics segments for AI-referred traffic.
Days 31 to 60: Connect and enable
- Automate data flow between your ERP, PIM, inventory and store.
- Enable the AI and agentic channels your platform supports, starting with a subset of products.
- Improve landing pages and checkout for high-intent AI visitors.
Days 61 to 90: Extend and measure
- Review AI traffic, conversion and revenue against other channels.
- Pilot an on-site AI shopping assistant for one category or for order support.
- Plan API work for UCP, ACP or MCP-based integrations if you run a custom store.
At the end of 90 days you will know how much AI traffic you receive, how well it converts, and where further investment will pay off.
Frequently Asked Questions
What is agentic commerce?
Agentic commerce is online shopping in which an AI agent acts for the shopper: it understands their request, searches and compares products across stores, answers questions and helps complete the purchase, while the merchant usually remains the seller of record.
What is the difference between UCP and ACP?
The Universal Commerce Protocol (UCP) was launched by Google in January 2026 and covers the full shopping journey, powering checkout in Google AI Mode and Gemini. The Agentic Commerce Protocol (ACP) was released by OpenAI and Stripe in September 2025 and defines how agents and merchants exchange product, checkout and payment information. Both are open standards.
Can customers buy directly inside ChatGPT?
OpenAI launched Instant Checkout in September 2025, but in March 2026 said it was moving Instant Checkout into apps and prioritising product discovery. ChatGPT remains an important place where shoppers discover products, and merchants can share product data through ACP.
Do I need to rebuild my store for agentic commerce?
Usually not. Most stores start by improving product data, feeds and structured data and by enabling channels their platform supports. Custom or headless stores may need API work to support agent integrations.
Will agentic commerce replace my website?
No. AI agents are a new discovery and buying channel, but your site remains where many purchases complete, where you control the brand experience and where you build direct customer relationships.
How do I know if AI assistants are sending me customers?
Segment referral traffic from AI assistants in your analytics tool and compare its conversion rate and revenue per visit with other channels. Many analytics platforms can now identify common AI referrers.
Is agentic commerce safe for payments?
The main protocols use scoped payment tokens so agents do not see card details, and the merchant stays responsible for processing. You still need fraud controls, clear policies and monitoring for agent-initiated orders.
How can CodeBase Coders help my store with agentic commerce?
CodeBase Coders gets your store ready for AI shopping agents end to end. We audit and clean your product data, build e-commerce and storefront changes, connect your catalogue, inventory and order systems through API integration, and implement the protocols your platforms support. We also build AI shopping assistants for your own site and AI agents that automate catalogue and order work, and integrate AI into the systems you already run. Book a free agentic commerce readiness review.
Sources
- Google: Universal Commerce Protocol announcement (11 January 2026)
- Stripe: Stripe powers Instant Checkout in ChatGPT and releases the Agentic Commerce Protocol (29 September 2025)
- OpenAI: Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol
- Digital Commerce 360: OpenAI shifts checkout plans in its agentic commerce strategy (6 March 2026)
- Microsoft Advertising: Conversations that convert: Copilot Checkout and Brand Agents (8 January 2026)
- Digital Commerce 360: Generative AI shifts online holiday shopping traffic in 2025 (Adobe Analytics data)
- Digital Commerce 360: Adobe AI-referral traffic data for July 2026 (19 August 2026)
Work With CodeBase Coders
Agentic commerce rewards stores that are easy for AI to understand and trust. CodeBase Coders helps businesses turn ideas into scalable digital products and improve existing processes through software, automation, AI, integrations and modern web technologies. We can clean and connect your product data, upgrade your store, and build the AI assistants and integrations that put your products in front of AI shoppers.
- E-commerce Development
- Software & API Integration
- AI Integration Services
- AI Shopping Assistants & Chatbots
- AI Agent Development
- Web Application Development
Ready to be recommended by AI shopping agents? Book a free agentic commerce readiness review with CodeBase Coders and we will check your product data, feeds and systems and give you a prioritised plan. Explore everything we build at codebasecoders.com.
Rohan Verma
Founder, CodeBase CodersRohan Verma is the founder of CodeBase Coders. He helps startups, SMEs and enterprises turn ideas into scalable digital products and improve business processes through custom software, AI, automation, integrations and modern web technologies.