One of the things I like the best about ecommerce is that it is always evolving, based on consumer behavior and new technologies. One of the best examples of this change is product discovery: We went from typing a shop URL directly, to searching for products on Google, searching them on Amazon, then on Tiktok. Is the next step to have ChatGPT become our shopping concierge, doing all of the heavy lifting for us?
With OpenAI becoming a for-profit company, ecommerce is no longer just a distant option, it’s a clear target for AI platforms. Over a couple of recent LinkedIn posts, I discussed how AI assistants could sidestep Google and directly show consumers products they can purchase, right within third-party apps, earning a fee on the sale.
That raises an important question: would these agents act in customers’ best interests, or in service of their own bottom line? This would not be just a minor shift, it would be a huge change in product discovery and purchasing, a trend already known as agentic commerce. In this article, we will discuss what agentic commerce really means, how it challenges, but also creates opportunities for brands with a DTC presence, and what the future might look like as platforms like Shopify, OpenAI, and Amazon invest massively in this technology.
What is Agentic Commerce, and Why It Matters
Agentic commerce refers to AI agents that do more than just giving product recommendations and answering questions, they search for products, evaluate them, and even complete purchases on behalf of customers.
Instead of clicking on a million search results or wasting time navigating brand sites, online shoppers can give quick AI agents instructions (“find me hiking shoes under $100,” “find the best budget earbuds available now”), and the agent researches, compares, and even buys for them. A recent article from Brian Warmoth describes agentic commerce as “a layer of software built on top of large language models… designed not only to interact with humans … but also to make decisions and execute actions autonomously”.
The consequences for brands are massive. Bernstein estimates that agent-driven experiences could increase global ecommerce conversions by 1.5 to 2.5% annually, which represents more than $240 billion in extra retail revenue. Walmart has already set up “super agents,” OpenAI is testing built-in checkout features in ChatGPT. These agents aren’t just futuristic projects, they are real and are driving a lot of investment across the industry.
How Agentic Commerce Will Cause Disruption and Opportunities
This new way of shopping raises questions for brands, especially for those relying mostly on D2C channels. First, how will agents find products to recommend?
In this new era, brands would need to sell not only to customers, but also convince AI-agents their products are worthy of being shown to customers. Traditional websites, UX storytelling, and flashy social media presence may not resonate with AI agents that prefer hard data: product descriptions, price, reviews, ratings, or API availability. This means brands will need to create two parallel experiences, one for real customers and one for AI-agents. The other implication is AI-agents will search for the best deal for each product, most likely favoring the original brands, which will hurt dropshippers and resellers.
Another problem for brands is the loss of human traffic on their D2C channels. A healthy D2C channel isn’t just about margins, it’s about building customer connections, and data acquisition. Initial CAC for D2C channel is often higher than let’s say a large marketplace, but repeat customer financials often beat marketplace margins and fees. And let’s not forget that data and feedback from direct customers can be invaluable for brands, while marketplaces policies tend to voluntarily limit interactions between brands and customers.
But it isn’t all bad news and some brands may have a lot to win. I personally think it could have a moderate impact on revenues in the next few years, but others are way more optimistic: Grid Dynamics reports agentic commerce brings massive gains: +30% cart conversions, –50% customer support costs, and +40% faster order fulfillment
Can We Really Trust AI Agents?
With agents acting autonomously, trust becomes even more important in ecommerce. It’s not enough for AI to make the buying process easier, it must also feel safe, transparent, and auditable.
The first question that came to mind when I first heard of agentic ecommerce was, how will agents pick the products to show the customer? Will they act in the customers’ best interest, or in their own interest? And how do we ensure the cards are not stacked against consumers? If a customer searches for an entry level camera, they don’t need to be recommended high-end models that may earn the platform a larger commission.
Consumers are concerned, too. In a recent Omnisend survey, 66% still prefer to make purchases themselves rather than giving control to AI agents; only about 34% would allow AI to buy on their behalf. Privacy and recommendation accuracy were top concerns. Brands and platforms must design for transparency, consent, and user control, not just efficiency. But the amount of money involved may grow so large that we must keep in mind the risks involved.
Regulation will follow, but will not be easy to set up, especially knowing how fast the environment is changing. Sellers need to prepare for policies around agentic commerce, fee disclosures, fairness across brands, and even algorithmic bias in product selection. Will some competitors try to game the system? Most likely. But ignoring them won’t help, and staying up to date with current trends is key.
Conclusion
Agentic commerce is no longer sci-fi, it is taking more and more space in the ecommerce world. Platforms like OpenAI, Google or Amazon are racing to get ahead, building the architecture that will let agents act autonomously on behalf of customers. For brands, especially D2C players, this means both risk and opportunity.
Success will come to those who adapt product data, APIs, and bot-friendly catalogs, without forgetting about the real humans who shop the brand. But brands must also understand transparency is key, and will need to figure out how agents pick the products to be recommended to customers.
In the near future, we may see a shift from “shopping UX” to “robot-friendly UX,” from “marketing funnels” to “agentic signals,” and from controlling the customer journey to influencing the AI-agent journey.
Is this the end of D2C? No, but it is a significant change, and brands must prepare for it.
