Agentic AI, where autonomous agents perform actions on behalf of people, may be relatively new, but it’s advancing at a rapid pace. In ecommerce, we’re beginning to see these agents built and deployed to browse and buy products for people. This technology is so transformative that we may be on the edge of a new era of ecommerce.
Unlike Alexa’s early promise of a new way to shop, this evolution is happening within a visual medium already familiar to consumers: websites. But are consumers ready for it? A survey published by Omnisend showed that 66% of consumers refuse to let AI make purchases for them, even if it promises better deals.
While this is likely still true for the majority, the tide is changing. Remember, adoption takes time, and as users become more familiar with the tools these agents are built on, like ChatGPT, adoption can escalate quickly.
Let’s look at what’s already familiar to consumers.
Advertisement
Amazon. Its Buy For Me feature, released in beta in April, allows customers to purchase products autonomously from other websites while still on Amazon. While Buy For Me still requires human involvement up to the point of purchase, it’s only the beginning.
Walmart and Mastercard. Mastercard released its Agent Pay feature while Walmart is developing an agentic AI tool to help shoppers navigate the purchase journey, including “shopping journey completion.”
ChatGPT and Perplexity Shopping features are designed for users to conduct research on products and discover the “best” product options. Users can interact with the tool and refine results based on requests.
These are all companies that are well-known to consumers. But here’s where it gets interesting from an ecommerce standpoint.
OpenAI’s Operator combines its own browser into the request process. With Operator, a user can prompt the tool to search for a product using as many defined criteria as they like and watch as it autonomously navigates websites on the same screen, completing tasks, including making purchases. This is a major next step in the evolution, and we can all see where this is heading.
And then there’s Google. Its AI Mode is the closest yet to a true agentic AI shopping experience. Users can virtually try on products using their own likenesses, track and set purchase prices and make purchases themselves. Considering Google’s importance to the current shopping environment, this is massive.
So how long before Target, Apple, Meta and other AI platforms take the next step in their already-developing AI commerce solutions?
Consumers may not be ready just yet, but that doesn’t mean they won’t be, and it isn’t stopping companies from going all in on it. They want to be ready the moment consumers are, and even before.
This is the race companies want to win.
But What Does Winning Look Like?
This is what concerns me most. Are we looking at an eventual consolidated ecosystem? If so, will it be a good thing, and for whom?
As we see, there is no shortage of companies trying to own the AI shopping ecosystem, but how will this affect consumers? What happens if one AI agent rules them all? Who then has the power, the retailer or the AI?
Right now, it’s looking like a race to own the ecosystem. When we talk about owning the ecosystem, we talk about the biggest players on the market, including Google, Amazon, Walmart, widely adopted AI platforms and even financial institutions.
But here’s the rub: if you own the ecosystem, you own the data. If the AI agent is controlled by the same company that owns the marketplace, the data layer, the payment stack and possibly even the fulfillment network, competition becomes almost impossible. It may no longer deliver the best results for the query, “What’s the best size 13 running shoe for a casual runner?” Instead, it may deliver results based on what’s best for the agent’s creator.
Will a closed system like this reduce shoppers’ exposure to products simply because an ecommerce company didn’t pay to be included in the results? Or will it return results based on volumes of available content, and if so, will this favor large companies like Walmart?
I pondered this same consolidated ecosystem back in 2017 when Alexa was dominating headlines. I thought we might be looking at a future where companies like Apple, Google and Amazon tried to “own” homes — WiFi networks, thermostats, home security, internet browsers, entertainment channels and smart devices. It appears this is where shopping AI agents are heading.
If the market consolidates, I can’t see how it helps ecommerce brands or shoppers. Will small brands be able to compete?
Would an Open Agentic AI Ecosystem be More Beneficial?
Generally, giving consumers a choice is a good thing, but would that be the case with AI agents in ecommerce? If the marketplace were to remain fragmented, made up of both large and small companies, consumers would need to choose their preferred platform. This invites a slew of questions.
Would all agents be able to operate across all websites and platforms? If not, would consumers need to use multiple agents to execute different tasks? How will they decide which one to use, and for what purposes — after all, you can’t have two agents looking to purchase the same products at the same time. Will one agent feature different product and brand results than another?
This seems to create more friction in the shopping journey than less.
We can’t talk about online shopping without talking about discounts. How will shoppers know whether or not they got a good deal on a product? Sure, an agent can monitor prices and report that the price is lower than normal, but this would presumably be for the listed price. Will the agent be able to apply discounts, such as those received for signing up for an email program, that would bring a higher listed price lower than the less expensive price?
If these questions can’t be adequately addressed, a fragmented ecosystem might encourage consumers to abandon shopping AI agents altogether, If this happens, AI shopping agents might transform into product discovery and deal-finder tools rather than transactional ones, leaving the current shopping process altered but mostly the same.
There are so many unknowns.
Looking Ahead
At the end of the day, it’ll be fascinating to watch how this all shapes up. Agentic AI has the potential to shift how consumers interact with brands. Consumers may have brand loyalty but agents won’t, unless, presumably, directed by individuals to favor those brands. If a shopper directs an agent to purchase a product and references brands whose products they typically like, this is what the agent will find. If you’re a lesser-known brand, how will you be discovered by the agent (unless, of course, it’s pay-to-play)?
We may need to redefine customer loyalty, from the traditional definition to simply being a “preferred option.” This means building a brand will be more important than ever, causing brands to focus more on the post-purchase experience than they do now, which is primarily an acquisition-first approach. The downstream effects may heighten the importance of first-party channels like email and SMS for post-purchase and product discovery communications.
The agentic AI revolution in ecommerce is happening, and brands need to think about how it will impact them. My advice: start now.
Greg Zakowicz is Ecommerce and Retail Advisor for Omnisend, an email and SMS marketing automation platform. With nearly 20 years of experience as a practitioner, consultant and analyst, Zakowicz has helped countless brands optimize their digital marketing strategies. He’s a frequent speaker at industry events, a published writer on ecommerce trends, has been featured in top-tier media outlets and has served as an industry expert witness for trial.
