Follow ZDNET: Add us as a preferred source<!–> on Google.
Our friendly neighborhood AIs are now able to do a new trick. As ZDNET’s Elyse Betters Picaro has shown, AI-enabled browsers like OpenAI’s Atlas and Microsoft’s Edge can do some of our shopping for us.
While AI-enabled shopping might seem like a gimmick akin to showing off robots dancing on America’s Got Talent, agentic commerce (which is what the wonks call it) looks like it has some legs.
What does it mean for consumers? What does it mean for the big money interests? And what does it mean for small businesses, which must navigate all these changes to stay competitive?
Stay tuned: I have seven important tips that can help small business owners stay ahead of the changes and thrive in this new environment.
Let’s dig in.
Online shopping isn’t much fun anymore
Online shopping has gotten messier and more spam-filled. According to Amazon, the company “blocked over 275 million suspected fake reviews from its store.” Obviously, Amazon is taking a proactive approach, but consumers are still finding enormous amounts of review slop within Amazon’s listings.
Also: PayPal’s new service is built for the AI shopping future – security included
And while in-person retail still outpaces online shopping, it doesn’t mean consumers are fleeing Amazon. According to JP Morgan, as reported in Yahoo Finance, Amazon owns 46.8% of the US e-commerce market. That market share has been growing year-over-year. Basically, half of all online shopping is Amazon. Everyone else has to fight for the other half.
special feature
All of this leads to choice overload. There are too many listings, too many sponsored placements, too many fake reviews. Consumers, pressed for time, are gravitating toward shopping methods that reduce effort.
One clear example of this is the growth in “buy again” subscriptions and automated purchases. According to Fortune Business Insights, the global subscription e-commerce market grew from $2 billion in 2024 to $2.7 billion in 2025. By 2032, that number is expected to explode to nearly $7 billion.
My wife and I are a prime example of the subscription model. We have a fairly extensive laundry list of Subscribe & Save items. These include actual laundry supplies, dishwasher detergent, toilet paper, healthcare items, and a vast list of other consumables that we use and replenish on a regular basis. Although we regularly check on and manage what’s going to be sent next (for example, we sometimes delay a given item by a month or two if we’re going through it more slowly than normal), it all happens mostly automatically.
The dawn of agentic commerce
Consumer ease (and the rampant desire for additional monetization opportunities) is what’s driving the rise of AI shopping agents. Effectively, this transition involves shifting from searching and clicking to delegating and making decisions.
Mastercard’s chief AI & data officer, Greg Ulrich, told ZDNET, “Agentic commerce is a new form of online and mobile shopping, in which an AI agent completes tasks for a user, such as searching for items, comparing options, and making a purchase, with limited or no manual inputs needed from that person or user.”
Also: Ready for AI-enhanced credit cards? Here’s Visa’s vision of automated shopping
Today, consumers search, compare, read reviews, check prices, and then make purchases. The idea behind agentic e-commerce is that consumers simply have to express intent. Then the AI agents search, compare, decide, initiate purchases, and track.
Clearly, all of this is happening at a level well beyond the scope of the typical small business. So what does all of this mean to small businesses? How can small businesses survive and thrive in this coming environment?
7 things a small business can do
To survive in an agentic shopping era, you’re not only going to need to make customers aware – and fans – of your products and services, but you’ll also have to reach the AI agents themselves.
Essentially, you’ll have to do influencer marketing to robots. In practice, this means you’ll have to provide product information online that’s structured and understandable by AI agents. That way, you can be discoverable to them where they’re scanning for product details.
The idea of this, along with the wide prevalence of standards and formats, can be overwhelming, but some of it is actually not all that difficult. Most e-commerce providers – such as Amazon, Shopify, Square, and WooCommerce – offer listing and metadata tools. Google does as well.
Also: How to actually use AI in a small business: 10 lessons from the trenches
The deep details go beyond the scope of this article, but here are seven key takeaways that should get you started.
1. Follow data format rules: Fundamentally, it comes down to the idea that if a commerce provider wants data in a certain format, or wants a certain form filled in, it’s absolutely in your best interest to do it.
2. Provide complete product metadata: Make sure you use proper GTIN/UPC codes, assign unique SKUs to all your products, and provide metadata that includes unique product titles, descriptions, prices, dimensions, variants, shipping availability, brand, stock status, and more.
3. Use APIs for real-time updates: Look for APIs required by or provided by your distribution channels. You’ll need to be able to update pricing, availability, shipping timelines, and return policies in real time. Keep in mind that AIs will constantly rate your data for quality.
4. Bad data hurts AI trust: If you indicate something is in stock when it isn’t, and an AI places an order that does not succeed, the AI agent will de-weight your offerings to reduce the chance of it happening again. This could be a challenging time, as there may be no way to appeal such an action, other than for AIs themselves to decide when to purge their old ratings and refresh.
5. AIs will judge your business reliability: As a small business owner, you’ll need to keep up with your existing channels. However, your digital messaging will be read, parsed, and either integrated or discarded by AI agents that, as part of their mission to shop, look for consistent and reliable data.
Also: I run a very small business. Here are 21 simple ways AI saves me time every day
6. Make sure information is consistent everywhere: Beyond all of the APIs and listing forms, make sure the rest of your information is consistent across your various social media feeds, your various store and marketplace listings, the various third-party aggregators and marketplaces that carry or list your items, in manufacturer databases that catalog your items, and even in downloadable documents that mention your offerings.
7. Think of it as AI SEO: While it’s certainly reasonable to provide different offerings to different outlets, the most important thing you can do is present your information to our new robot overlords in a way that makes your business appear reliable. Think of the new normal as SEO for AI.
The consumer trust issue
In theory, agentic AI will help insulate consumers from ads, irrelevant products, counterfeit listings, and review spam. AI agents would automatically optimize loyalty programs, track return policies, and perform real-time, constantly updated analysis on behalf of harried shoppers.
But this change won’t be absolute. Most likely, consumers will delegate some purchasing tasks (like my wife and I do with Subscribe & Save) but stay hands-on with others. I recently purchased a winch (because of course I did). The product choice search process was entirely hands-on, although I did turn to ChatGPT to help me understand the scope load calculations, which I externally confirmed using other tools
The hype behind the safe growth of AI shopping agents is that, rather than interacting with e-commerce interfaces, consumers will interact with the AIs, delegating the work of selecting, comparing, and purchasing products.
You can bet that AI agents will also provide spammers and scammers with a new entry point for gaming the system. We’re already seeing how millions of Russian propaganda articles have been posted with the specific goal of influencing AI systems and the responses AIs give.
If AIs are given the authority to spend actual money on behalf of consumers, can there be any doubt that entire industries will be devoted to influencing, optimizing, and even hijacking that money flow?
Also: People don’t trust AI but they’re increasingly using it anyway
Mastercard’s Ulrich insists that his company’s AI transactions will execute under “verified intent,” to provide a measure of trust and permission. Each transaction must be traceable, revocable, and operate only with user permission.
The company is introducing a service called Agent Pay, which Ulrich says is “a trust layer for agentic commerce. It empowers AI agents to act securely and autonomously, while giving consumers and businesses full control over what gets purchased, when, and how.”
Big money interests
Of course, none of this is truly about the consumer. It’s about the big money interests and how they can produce more output with less cash and make more money at a lower cost.
It also isn’t just about storefronts and e-commerce. It’s about a lot more than that. For one thing, it’s about who owns the infrastructure AI agents use to shop and pay.
For payment companies like Mastercard and Visa, it’s all about payments, identity, and tokenization (who owns the transaction). For big finance suppliers like Oracle and SAP, it’s about supply chain, finance, and procurement automation.
For traditional players like Amazon, Google, and Apple, it’s all about the consumer interface, protecting their turf, and reinforcing their ecosystems. And for the big AI players like OpenAI, Microsoft, and Anthropic (which has a major partnership with Amazon), it’s about disrupting the old guard and becoming the new front of commerce overall.
In an interview with ZDNET, Hari Sankar, group vice president of product management at Oracle, said, “Agentic AI embedded directly into applications allows business leaders and finance teams to bring the AI to the data, not the other way around. This can allow AI to leverage contextually relevant and up-to-date data and enables AI to take immediate action.”
Look at that statement carefully. What he’s essentially saying is that AI power will belong to whoever owns the infrastructure layer where the data lives. Agentic AI for e-commerce isn’t really about consumer enablement. It’s about owning the deep infrastructure.
Sankar contends, “Agentic AI is only as good as the data it uses and the processes it orchestrates.”
In other words, control of data and process pipelines, such as finance systems, supply chains, and procurement, is what really matters. Websites and storefronts are completely secondary.
This infrastructure theme is echoed by Mastercard’s AI chief. “We’re very intentionally building agentic commerce so that each transaction made is permissioned, with digital credentials verifying what’s being purchased, from whom, and for how much.”
The future of e-commerce
The folklore is that in-store retail shopping has declined. But that’s not really true. Except for 2020, in-store retail shopping has actually grown year-over-year. It’s just that it didn’t grow all that much, and online shopping has been growing by double digit percentages each year.
Still, traditional shopping malls are hurting. As of the fourth quarter of 2024, Capital One reports that “the nationwide mall vacancy rate is 248% higher than the overall average retail vacancy rate.” The credit card issuer also reports that “up to 87% of large shopping malls may close over 10 years.”
Those half-empty malls didn’t die overnight. Their structure just didn’t evolve with consumers’ changing needs and desires. They also couldn’t compete with new business models, such as those of the giant home centers, which were designed specifically to remove mall landlords from the equation. Initially, this was a way to control costs, and later, it became a means to grow without the limitations of the mall formula.
Also: The best e-commerce website builders: Expert tested
For the past decade, business advisors have touted digital transformation and multi-channel selling, the idea that companies need to meet customers wherever they happen to be, typically on their smartphones.
But now, businesses will need to meet their customers’ bots wherever they happen to be, and that’s likely via APIs and clear, consistent data. If the machines can’t find your information, you’ll be left behind.
Online vendors that ignore or underutilize AI-enabled commerce may well suffer the same fate as shopping malls. Once again, an outside business model is providing additional options for growth by changing how customers are served.
Rather than thinking of the various AI agents as customers, think of them as procurement officers, as purchasing departments not only for companies, but also for individuals who rely on AI agents for purchasing. If your products can be found, parsed, trusted, and bought programmatically, you’ll be well-prepared for the new agentic adventure we’re all about to embark on.
What about you? Do you think you’d trust an AI agent to shop or make purchases on your behalf? Have you already used tools like Subscribe & Save, or are you still more comfortable making every buying decision yourself? If you run or work for a small business, are you preparing for this shift toward AI-driven commerce, or does it feel out of reach? And do you think this new model benefits consumers, or only the big companies building the infrastructure? Let us know in the comments below.
You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.
