From ChatGPT to shopping agents: The silent disruption already under way

For anyone who’s used ChatGPT before, that won’t feel like a big leap. It’s just the natural next step in a conversation about products.

But it’s also something more. It’s one of the first public signs of a much bigger shift and a move toward agent-based shopping, where AI systems act on behalf of the user to discover, evaluate, and even buy things. If you work in retail, brand, or mobile experience design, this is worth paying attention to sooner rather than later.

From search to assistance

This isn’t search in the traditional sense. The difference isn’t the interface — it’s the intent. You’re not scanning a list of links or ads. You’re having a conversation. You ask a question, the assistant responds, ideally with suggestions that match your context, your taste, and your needs.

Right now this lives inside ChatGPT. But we all know it won’t stay there. As large models get integrated into phones, apps, operating systems and messaging platforms, the act of shopping will move upstream into conversations, voice commands, reminders, even background nudges. The assistant becomes the storefront and the phone becomes the terrain.

The implications for mobile are significant. Today, the current version mostly links out to mobile web. But the next version won’t stop there. Agents will need to deep link into apps, open product pages, trigger service flows, and pass through context — all without breaking the conversational thread. If your mobile app doesn’t support that kind of interaction, you’re at risk of being bypassed completely.

Assistants that learn from you

These new agents won’t just answer questions. They’ll remember what you’ve asked before. What you’ve bought. The brands you like. The ones you avoid. Your delivery constraints, your budget, your loyalty points and even how you prefer to pay.

They’ll be able to weigh those things in context. Maybe one retailer is a few pounds more expensive, but you’re part of their loyalty scheme. Maybe another offers faster delivery, or aligns better with your sustainability values. A third might let you use your preferred payment method. These aren't things traditional filters handle well, but to a context-aware assistant, they’re all part of the equation.

Over time, users will effectively be training their agents, even if they don’t realise it. Not with settings menus, but with habits and preferences — what they choose, what they scroll past, what they ask again. That kind of memory gives the user real control.

And it’s the kind of control that’s been missing from most digital experiences. Want to stop seeing baby products now that your kid’s in school? Want to shift from fast fashion to sustainable brands? Want running gear because you’ve just started the Couch to 5K? Just say so. That’s the promise: shopping that adapts to you, not the other way around.

That changes how trust works too. If an assistant recommends a brand you’ve never heard of, what makes you click? It’s no longer about billboards, or logos, or ad impressions. It’s about signals of credibility; reviews, reputation, ratings, and what other people like you have chosen. Which means brands may need to rethink how they build awareness in a world where assistants are the gatekeepers.

What this means for the current ecosystem

Let’s talk about who feels this shift first.

SEO: If agents are handling discovery, traditional SEO starts to wobble. We’ll see a shift from optimising pages to structuring data so it’s readable, parseable, rankable by an agent, not just a search crawler. Think less “first page of Google” and more “first suggestion from your assistant.”

Affiliate and aggregator sites: These are already starting to look exposed. If ChatGPT or similar agents can summarise Reddit threads, pull in product specs, and give direct links — then the value prop of “comparison” sites starts to wear thin.

Amazon and big marketplaces: This one’s interesting. Amazon still has scale, logistics, and trust. But if I’m using an agent that searches across retailers, price-matches automatically, and isn’t biased toward any one platform, then Amazon becomes one option, not the default. That weakens their grip on discovery, even if they still dominate fulfilment.

And underneath all of this is a bigger point: if the agent is now mediating the relationship, brands can’t afford to lose direct connection with their customers. Being visible in the app store or the search engine isn’t enough. You need to be relevant, trusted, and available to the systems people are actually using.

So what happens next?

If agents are here to stay (and it’s looking that way) then a few new roles are going to emerge.

Product data translators: Brands will need to make their product info agent-friendly, not just web-friendly. That means structured data, good metadata, clear attributes, stock info, delivery options — all in a form an AI can reason with.

Reputation layers: Agents need help assessing quality. That could mean trust signals, verified reviews, or APIs that summarise sentiment. There’s space here for new intermediaries, not for users, but for the agents themselves.

Preference brokers: If users are going to start training their agents, implicitly or explicitly, there’s a case for managing that data in one place. A personal profile, portable across agents and platforms, that holds your values, your needs, your red lines. Done right, it’s a new kind of asset - your shopping DNA.

And for mobile-first brands, this is where it gets real. Mobile is where these agents will live. It’s where they’ll surface suggestions, trigger flows, and follow through on intent. If your mobile experience isn’t set up to respond to agents, to listen and not just broadcast, you’ll miss the moment.

A final thought

This shopping feature in ChatGPT isn’t the endgame. It’s a signal. A test balloon. But it shows us where things are headed: toward a world where buying decisions happen earlier, faster, and with fewer clicks — because someone else is doing the thinking for you.

If you’re building for that world, the rules are changing. You don’t need to be louder. You need to be present, useful, and aligned with what your customer actually wants, when and where they want it. That starts now.

Need a chat to run through your app and see if you’re AI Agent ready? Reach out my AI Agent Preparation team to make sure you are ready for what’s coming.