The End of Browsers: The E-Commerce frontend is Dead

(…or at least getting compressed into a 30-second chat)

This is part of my series on how AI will probably rewire teams, careers, and GTM. Today: e-commerce. It’s a forecast exercise, so do reach out if you have a different view on the future!

We are all seeing the shift in traffic to ChatGPT, and starting with consumers. Here’s my take on how it’s going to play out:
We’re currently standing at about 10% of traffic with LLMs of one kind or another, with ChatGPT obviously being the forerunner.
40-50% of consumers say they’re already using gen-AI for shopping tasks (research, recs, deals). Meanwhile, when Google shows an AI summary, users click less. That’s demand capture moving from websites to answers.

The Monetization Play (and why it touches your P&L)

LLM platforms don’t have Google’s AdWords print-money machine. They are losing money, and badly so.
They need revenue, and aside from subscription, the next logical place is transaction revenue. Reporting indicates OpenAI is building checkout inside ChatGPT with a commission model. Perplexity already switched it on in the U.S. with PayPal/Venmo.
I started sharing that thought in podcasts in June/July, reality has caught up pretty fast!

That monetization pressure will have an impact on the e-commerce industry, and this is its possible future.

Take a jewellery online shop.
Historically: Google → ads → 6 tabs → Front End Ecommerce → cart.

It’s a high-friction, low-conversion storyline (only 1 in N ecommerce stores gets the conversion).

In an LLM world: “Need a modern necklace under $300, minimalist.”
The agent returns options with provenance and reviews.
Buyer: “#2 looks great! Let’s get it”.
Done. No category page. No checkout form. Low friction.

Would people let an AI place the order? Some will. Surveys showed ~34% are already willing (back in the ye old days of March 2025…).
It won’t be 100% of buyers; it doesn’t need to be. Even 20-30% of purchase volume shifting to agentic flows bends the stack.

Everyone assumes the future means ChatGPT shows the customer a carousel, then they click a link to complete the purchase.

I disagree. Why can’t they just say, “Option B looks great, order it for me”? The AI already knows the customer’s address and payment details.

I know I’d be happy to do that in many of my day-to-day purchases. I don’t really need to see the storefront if I trust the LLM.

If this happens, the entire concept of a marketing funnel disappears. Because e-commerce shops no longer engage with humans, they only interact with AI agents executing purchases.

And you have no sway over that agent.

The Monetization Play

Here’s how OpenAI actually makes money in this new business model: transaction fees on every purchase their AI completes.

You describe what you want, ChatGPT shows you options, and you pick one. For a small fee, it handles the entire purchase without friction.

This is OpenAI’s real monetization play. Not subscriptions or API calls. A 10% cut on every transaction that flows through their platform, just like the Apple app store (yes, Apple takes 30%, but margins on software are significantly higher).

I know that if I were working on monetization at OpenAI, I’d go all in on that project.

The B2B opportunity would work the same way. Imagine you’re a marketer at Cursor who needs contact data for 50,000 CTOs at the top engineering companies in the US. Instead of logging into ZoomInfo, navigating their interface, building lists, and managing credits, you just tell ChatGPT: “Pull me 50,000 CTO contacts from companies with the most engineers.”

The AI connects to ZoomInfo through Model Context Protocol (MCP), charges you $9, and delivers a clean CSV.

The Death of Digital Marketing as We Know It

When AI agents become the primary purchasing interface, the marketing stack you’ve spent years building becomes worthless. I’m not talking about optimizing for a new channel; I’m talking about the complete elimination of channels as a concept.

Channels are avenues to engage with your prospective audience wherever they might be.

Of course, they’ll still be on Facebook and on TikTok, and you can influence their desires over there.

When people stop opening browsers, every channel you’ve built your business on becomes irrelevant overnight.

No more searches because customers aren’t Googling products, they’re describing needs to ChatGPT. No more Google Ads because there’s no search results page. No more retargeting because there are no website visits to track. No more email newsletters because the AI agent doesn’t have an inbox.

The entire concept of a funnel disappears. There’s no awareness stage, no consideration, no decision journey. The customer describes what they want, the AI presents options, and they pick one. Done.

The shopping experience gets compressed into a single chat interaction, with no product pages, cart abandonment, or checkout optimization. The entire “customer journey” becomes a 30-second conversation.

And if that happens, 90% of today’s e-commerce and B2B marketing stack will be worthless.

Resistance is Futile

Not everyone will adopt this immediately. Like every major change, resistance comes from two places: fear and preference. This one will be no different.

A large part of the population will resist putting their credit card details in the hands of a semi-autonomous AI agent. That’s a reasonable fear.

But some of us will find ways to mitigate the risk. Give the AI a locked $500 card. Set spending limits. Reduce exposure while capturing the convenience. The early adopters will figure out the safety mechanisms.

The other argument I hear is that people enjoy the shopping experience. The CMO of Klaviyo told me that consumers like browsing websites because it’s a virtual form of window shopping. I know many people do.

There is an alternative future

I’m a big fan of dystopian futures, but truth to be told, there is an alternative possible future that is likely. Thanks to Noah (check out his Substack

the gtm engineer

) for sharing his thoughts.

A different path would be that a subset of brands with deep first-party data and real loyalty keep purchases on-site by shipping best-in-class embedded LLMs.
Think Nike: fit/size history, returns, inventory, and a style graph power a concierge on nike.com/app that beats a generalist model and ends in one-click buy.

No detour through a third-party agent. If this plays out, it becomes a race to operationalize proprietary data fast, likely accelerating brand consolidation around those who can.

B2B mirrors it: “agentic websites” replace SDR demos; you interrogate the product, spin up synthetic demos, and get an implementation plan in chat.


Maybe that’s true for some part of the population. But there’s another part, like me, who don’t want to spend time shopping, especially not scrolling through endless Amazon options trying to figure out which seller is legitimate.

This creates a clear split. Traditional shoppers will continue browsing for discovery and enjoyment. Efficiency-focused buyers will adopt AI-mediated purchasing immediately.

The question isn’t whether everyone will switch. The question is: what percentage of purchase volume moves to AI agents? Even if it’s only 30%, that’s enough to end most of the traditional marketing infrastructure (VCs would walk away immediately).

To be transparent, there is yet another different possible outcome: Shopify and the rest of the stack build shopping agents that are really good and trusted by the buyers. I’m not holding out on that one, but I wouldn’t want to be against Tobi either.

=> These are possible futures:

• E-commerce front-end disappears. A few large consumer-facing LLMs take over, Shopify becomes a headless/ backend platform. That’s my bet.

• Large brands build out their conversational agents with first-party data. The experience is excellent. E-commerce shrinks to fewer fronts where this can happen (amazon + a few thousand great brands).

• The E-commerce stack can create something at scale that performs as well as 1 or 2 above. Shopify + the likes of Gorgias and Klaviyo enable this future.

The Strategic Question Every Marketer Should Ask

Are you preparing for a world where your customers never visit your website?

In an AI-first world, when that customer makes that purchase request through ChatGPT, the AI agent becomes the entire customer relationship.

Your E-commerce company still makes the sale, but you lose everything that comes after: the brand interaction, the data collection, the opportunity for retention marketing. You become a pure product supplier in a system where the AI controls the customer relationship.

The channels that stay relevant are those where there is trust left or where people spend time outside of LLM. Obviously, that’s going to be influencer marketing, TikTok, Facebook, and whatever comes up next. Word of mouth, referrals, and branding are probably going to be more important in the future than they have been up to now.

This is the future that’s emerging right now. The technology exists today. Consumer adoption curves and payment integrations are the only barriers left.

The companies that recognise this shift and adapt their strategies accordingly will thrive in the post-browser world. Those that don’t will find themselves competing for an ever-shrinking pool of traditional shoppers, while the majority of commerce moves to AI-mediated channels they can’t access or influence.

There’s a post-browser world looming over us. It’s no longer just zero-click internet, but a world where people no longer use browsers at all.

The question isn’t whether this future arrives; it’s whether you’ll be ready when it does.