Imagine your next customer never needs to visit your website.
Instead, she opens her personal AI assistant — say, in ChatGPT — and types: “I need a dress for a party tomorrow.”
The AI assistant doesn’t just suggest dresses. It evaluates, negotiates, and purchases the right one — entirely based on the user’s values and needs, as previously defined in the AI’s settings.
No ads. No clicks. No cart.
This is Agentic Commerce.
It sounds like science fiction, but it’s likely just around the corner.
OpenAI has launched Instant Checkout in ChatGPT (currently in the U.S.) and its own browser, ChatGPT Atlas. With the Agentic Commerce Protocol (ACP) and Google’s new Agent Payments Protocol (AP2), AI is beginning to take over what we’ve traditionally called the checkout phase of the customer journey.
So what does this shift mean for e-commerce players? How do you prepare for a world where AI agents communicate and buy on behalf of humans? And what becomes of your website and brand when ChatGPT and other AI models start facilitating the checkout process for you?
What Is Agentic Commerce?
Agentic Commerce is about AI agents — like ChatGPT, Google Assistant, or Alexa — making purchase decisions on behalf of consumers.
Instead of someone visiting your website or store, an AI assistant — a chatbot, if you will — handles it all. It remembers birthdays, chooses the right flowers for the right occasion (no more forgetting your anniversary bouquet!), and manages repeat purchases.
The agent acts on the customer’s behalf, based on their preferences, values, and past behavior.
Where today’s e-commerce focuses on attracting visitors and collecting their data, the next phase flips that dynamic: you’ll need to deliver data to customers and their AI agents so they’ll prefer you.
        
What Should E-Commerce Businesses Prepare For?
Data is — and will remain — king.
The winners of tomorrow will be those who master data quality, automation, and reliability.
“The main takeaway is that this AI revolution will require your data to be accurate. There won’t be a human double-checking mistakes anymore. If someone clicks an ad for a blue jacket in medium and lands on a product page showing an orange jacket in XL — that’s the end of the interaction,” says Thomas Halling, Solution Architect at Alpha Solutions.
“If your product feed isn’t precise — if color, size, or stock status don’t match — you’ll pay the price. AI will only buy from sources it can trust.”
That means data quality is everything — not just clean data, but machine-readable data that allows AI systems to make autonomous purchase decisions. The AI must be able to complete the order — not just send an email to the warehouse and hope someone responds.
If you promise something you can’t deliver, you’re out.
Five Key Focus Areas for Businesses
- Structured, machine-readable data feeds
 Product data, stock levels, shipping details, prices, images, descriptions, and reviews — everything must be accessible.
 
 
- Automated processes
 AI must be able to press the button itself. Your backend needs to be clear, streamlined, and reliable. You can promise a great level of service all you want, but if the AI encounters failure during checkout, it won’t come back.
 
 
- Open APIs and modern standards
 Agentic Commerce runs on open protocols. If your systems are locked behind proprietary walls, you’ll be left behind.
 
 
- Trust scores and reviews
 AI agents make decisions based on data and history — not gut feeling. Trustpilot scores and customer reviews will become central factors in how agents select supplier.
 
 AI agents will reward businesses that deliver as promised. That means customer reviews, delivery speed, and returns will directly feed into the agent’s decision logic. A store with high Trustpilot or Google ratings signals reliability and low risk — and AI interprets that as trustworthiness.
 
 Delivery, in particular, will become the most tangible indicator of reliability. If you promise next-day delivery but consistently ship late, you’ll drop in ranking. AI will favor stores that statistically deliver as promised.
 
 
- Security and identity management
 Security and identity validation will be crucial prerequisites for participating in Agentic Commerce.
 
 When AI agents transact on behalf of users, they must be able to log in, validate identity, and complete actions — without compromising business systems.
 
 “If AI agents are going to buy from you, you need to be sure it’s truly the customer’s agent — not just a random bot. That means using standards like Open Authentication (OAuth 2.0) to verify logins from Google, Apple, or Microsoft,” explains Thomas Halling.
 
 If you rely on outdated login structures or internal APIs designed only for your website, you expose yourself to risk once AI agents access your systems directly.
 
 “Many companies rely on ERP–webshop communication where security means closed networks and admin logins. That works fine for a webshop — but not in an agentic world,” Halling adds.
 
 Companies must therefore build robust identity infrastructures with authorized tokens, secure endpoints, and clearly defined permissions for what AI agents are allowed to do.
 
 It’s not about hiding data — it’s about protecting processes. Your backend should be transparent to AI — but only at the precisely defined touchpoints you control.
A New Marketing Model?
If checkout within AI interfaces becomes the norm, your website shifts from a lower-funnel conversion touchpoint to a top-funnel branding platform.
Instead of fighting for human clicks and attention, brands will need to make themselves visible and attractive to AI agents — the ones actually making the purchasing decisions.
“Google and OpenAI won’t abandon their business models,” says Thomas Halling. “They’ll just move the auction higher up in the conversation. You’ll pay for Gemini or ChatGPT to pick your store — not for a human to click your ad.”
The partnership between OpenAI and Microsoft reinforces this shift — it’s not just a technical alliance, but a strategic realignment of digital advertising.
Bing is now the built-in search engine for ChatGPT, meaning all search ads inside ChatGPT are served via Microsoft Advertising — not Google Ads. This is significant because OpenAI controls a massive user base, pushing Bing Ads into unprecedented exposure.
Halling also predicts future access fees for AI models to use large product catalogs — once data volume hits a certain threshold.
While no one knows the exact business model yet, one thing is certain: the computational power, storage, and data validation needed to run this at scale won’t be free.
Website as Top-Funnel: Will AI Replace the Emotional part of commerce?
When AI takes over the operational part of the buying journey, the role of your website changes dramatically.
Your website becomes your brand center — not a store in the traditional sense. It’s where inspiration and emotional connection happen. Customers must know, trust, and remember your brand — so that later, they can simply tell their AI agent: “Buy it from them.”
An AI agent won’t stumble upon your product through a successful TikTok campaign. It will find you because it systematically evaluates all options based on precise criteria.
Will marketing teams shift from building emotional connections to optimizing based on measurable datapoints that agents evaluate?
If so, that will redefine what performance marketing means.
Trust and Timeline: When Does This Become Real — and Can We Trust AI Checkout?
When will we actually hand purchase decisions over to AI? When will consumers trust ChatGPT, Google, or Amazon to make those decisions safely and correctly on their behalf?
“Some will jump in immediately — the same people who love that their Tesla drives itself and trust technology completely.
Others won’t — the ones who refuse to turn on lane assist in their car.
And then there’ll be an anti-trend — people who reject AI because of environmental concerns, once they realize how energy-intensive it is,” says Thomas Halling.
So when does it all become reality?
We’re already seeing the early signs.
AI assistants are starting to handle routine purchases. The infrastructure for Agentic Commerce is being built right now. OpenAI has rolled it out in the U.S. Google has launched its Agent Payments Protocol (AP2) with major partners across commerce, logistics, and payments.
Commerce platforms like Medusa are already releasing tutorials on how to integrate your webshop with ChatGPT.
In other words — it's on the move.
THOMAS HALLING
Solution Architect
 
                        