AI Browsers Are Changing E‑Commerce: An In-Depth Look with Live Examples

By Joshua Warren, CEO of Creatuity and Host of the Commerce Today Podcast

Hello! My name is Joshua Warren, and I’ve been working in e-commerce since the days of dial-up. As the founder and CEO of Creatuity (and host of the Commerce Today podcast), I’ve seen many waves of technology reshape online business. But nothing has excited me quite like the rise of AI-powered web browsers. In this article, I want to share – in plain English and first-person perspective – why I believe AI browsers are set to redefine e-commerce, and what steps e-commerce leaders (both B2C and B2B) should take right now to stay ahead.

The Browsing Revolution: From “Search & Click” to “Ask & Act”

Take a moment to think about how we use the web today. For decades, the routine has been: open a browser, type keywords into Google (or another search engine), scroll through results, click links, read, and manually take actions like filling forms or adding items to a cart. This “search and click” paradigm has been the status quo since the 1990s. AI browsers are turning that model on its head.

Instead of you doing all that work, an AI-powered browser lets you simply ask for what you want, and it can act on your behalf. It’s like having a super-smart assistant living inside your browser. Need to find a product? The AI can search multiple sites at once and show you the best answer directly. Need to fill out a form or even complete a purchase? The AI can often handle it, asking you for confirmation if needed. In short, these new browsers can search, summarize content, draft messages, fill forms, and yes – even buy things for you. This isn’t a distant future scenario; it’s already happening in 2025.

Why does this matter for e-commerce? Because how people discover and shop online is about to change dramatically. If consumers no longer have to click through search results or navigate your website’s menus, the traditional ideas of SEO, online ads, and user experience design will all need a rethink. E-commerce discovery and UX are being redefined by AI at the browser level. In fact, Sam Altman (the CEO of OpenAI) recently said that AI offers a “rare, once-a-decade opportunity to rethink what a browser can be”wired.com. It’s akin to the leap we saw when tabbed browsing first appeared – but potentially even bigger in impact.

What Exactly Is an “AI Browser”?

So what do we mean by an AI browser? Sometimes you’ll hear terms like “agentic browser” or “autonomous browser.” In simple terms, an AI browser is a web browser with a built-in conversational AI assistant. It merges the familiar interface of a browser (web pages, address bar, tabs) with the capabilities of an AI that you can chat with, give instructions to, and even delegate tasks to.

Instead of pointing and clicking, you navigate by asking. For example, rather than manually browsing Amazon for noise-cancelling headphones, you could just tell the browser, “Find me the best noise-cancelling headphones under $200 and add the top pick to my cart.” The AI might search review sites, check prices across stores, and then present you with one or a few options – or even place one in your shopping cart – all inside a single conversational interface. It’s like having a personal shopping assistant or research aide living in your browser.

One of the hallmark differences is that AI answers come first, links second. Traditional Google searches give you a list of blue links and ads, and maybe a snippet of info. AI-centric browsers flip this: you ask a question or give a task, and the browser’s AI gives you a direct answer or action immediately, often with sources or the option to dig deeper if you want. The AI’s answer is central; the list of websites becomes a secondary optionwired.com. This inversion has huge implications. Users might get what they need without ever clicking through to a website – a phenomenon sometimes dubbed “zero-click” search. In fact, early data shows that AI search tools have lower click-through rates to external sites because the AI summarizes so much information up frontibm.com. In other words, the AI browser might solve the user’s query within the browser itself, and the user never actually visits your site unless it’s necessary.

It’s worth noting that while these new AI browsers are making headlines, the majority of users today still rely on the big two: Google Chrome and Apple Safari. Chrome remains the most used browser globallywired.com. But the surge of innovation in AI browsers is a direct challenge to the status quo. Even the giants are reacting – Google is adding an AI chatbot (Gemini) into Chrome’s toolbar, Microsoft built an AI (Bing Chat) into Edge’s sidebar long ago, and other players are integrating AI features as fast as they can. We’re at the start of a new browser war, one where the battle is not over speed or privacy or extensions, but over who can provide the smartest, most helpful AI copilot for your web experience.

Meet the New AI Browsers (Comet, Dia, Atlas & More)

Let me introduce three of the leading AI-powered browsers you should know about, plus a few others on the horizon. Each of these has a slightly different angle, but all share the core idea of an AI assistant woven into your browsing. I’ve tested each of these personally, including live demos on the Commerce Today podcast, and I’ll share my impressions of how they work for real shopping tasks.

Comet by Perplexity – The Pioneer of Agentic Browsing

The first mover in this space was Comet, launched by the AI company Perplexity. Perplexity’s claim to fame has been its AI Q&A service (a bit like an AI search engine), and with Comet they extended that tech into a full browser. Earlier in 2025 Comet was invite-only, but it’s now available for anyone to download and use.

What Comet does: It acts like an autonomous agent that can execute multi-step tasks online. You can literally tell Comet to do something complex like “Book a table for two at an Italian restaurant this Saturday at 7pm, and email me the confirmation,” and watch it try to handle the whole workflow. It will search for restaurants, find a suitable one, attempt to make a reservation (using OpenTable or the restaurant’s site), and even draft the email. In testing, I had Comet add an item to an Amazon cart for me just by asking – it searched the product, clicked the right link, and put it in the cart. Comet combines web search intelligence with the ability to take actions in the browser (clicking buttons, filling text fields, etc.), all while remembering the context of your request.

Strengths and weaknesses: When Comet succeeds, it feels like magic. An IBM reviewer who tested Comet described watching it handle bookings and emails and said it “truly felt like the beginning of a new era in search”ibm.com. Routine tasks like finding a product, or checking a few different websites for the best price, become faster because Comet can do them in parallel and summarize the results for you. However, Comet is not perfect – it can get confused by unusual web layouts or make the occasional wrong click. In one case, I instructed it to reserve a table with a friend, and it accidentally emailed the wrong person because it misinterpreted my friend’s name in my contacts. These hiccups show that the tech is early. Still, Comet provided the first real glimpse of what a “next-era” browsing experience could be. It’s like a very eager intern: super fast at gathering info and taking initiative, but you still need to supervise to catch mistakes.

Dia by The Browser Company (Arc) – Your AI Shopping Concierge

Next up is Dia, launched by The Browser Company (the makers of the Arc browser). If Comet is like an intern that can do tasks for you, Dia is more like a smart assistant whispering in your ear as you browse. The Browser Company created Dia as a new AI-driven browser with the philosophy of being able to “chat with your tabs.” In practical terms, Dia doesn’t try to autonomously run off and complete multi-step jobs like Comet; instead, it’s embedded AI assistance at every turn.

What Dia does: Imagine you have 20 tabs open (we’ve all been there), and you’re trying to find which one had that shipping policy info. With Dia, you can just ask, “Which tab was talking about free returns?” and it will find it. Dia can summarize the content of any page you’re on, explain complex terms, translate foreign-language text on the page, or even help you draft a better email in your Gmail tab. It’s like a layer of intelligence on top of the web pages. For shopping specifically, Dia shines as a “retail concierge.” When you’re on a product page, you can ask Dia things like, “Is this a good price for this item?” or “What are the most common complaints about this product in the reviews?” and it will give you helpful answers inline. I’ve used Dia to sift through thousands of user reviews on Amazon – it can pull out the key pros and cons, so I don’t have to read 50 individual reviews. It will also let you know if the item you’re looking at is priced higher than on other sites or if a newer model is available. Essentially, Dia’s goal is to make sense of messy e-commerce data as you browse, pointing out comparisons and insights you might otherwise miss.

Why it’s noteworthy: Dia represents a slightly different approach from the others because it’s very user-driven. The AI doesn’t take over your session; it’s at your beck and call. This feels powerful in scenarios like research and shopping where you might not want the AI to go rogue, but you definitely want its help digesting information. A fun example – I was looking at a complex product spec sheet for a laptop, and I asked Dia to “Explain like I’m not a tech expert.” It produced a short, friendly summary of the key features in plain language. For e-commerce teams, this kind of AI assistance means customers could be better informed (and maybe less likely to misunderstand a product or miss an important detail), all without leaving your site. Update on Dia: If you follow tech news, you may have heard that in September 2025 Atlassian (the enterprise software giant) agreed to acquire The Browser Company – the creators of Arc and Dia – for a hefty $610 millionblog.brianbalfour.com. Atlassian’s plan is to transform Dia into “the browser for work,” packed with AI features for knowledge workers. This big acquisition underlines that major players see AI browsers as the next frontier. (Translation: this trend is real, and it’s not going away.)

OpenAI Atlas – The ChatGPT Browser That Can Do It All

The newest entrant (and perhaps the most ambitious) is OpenAI’s Atlas, which launched in late October 2025. Atlas is essentially OpenAI’s answer to Chrome, but built around ChatGPT. It’s a Chromium-based browser (meaning it feels a lot like Chrome in design) with ChatGPT built directly into a sidebar and deeper under the hood. I’ve been using Atlas since the day it launched, and it has quickly become my favorite because of how seamlessly it blends AI into everyday browsing.

Key features of Atlas:

• ChatGPT Sidebar: At any time, you can pop open a ChatGPT panel on the side of Atlas. It’s context-aware, meaning if you have a webpage open, you can ask ChatGPT questions about that page. Reading a dense article? Ask Atlas to summarize it. Viewing a product? Ask if there’s a cheaper alternative or what the specs mean. It’s like having the power of ChatGPT on-demand, without leaving the page. This feature alone is a productivity boost – no more copy-pasting URLs into ChatGPT; it’s all in one place.

• Agent Mode (for Plus/Pro users): If you subscribe to ChatGPT Plus or Enterprise, Atlas unlocks an Agent Mode. This is where things get wild – in Agent Mode, the AI can autonomously take actions in the browser just like Comet does. It can click links, scroll pages, fill out forms, and even navigate multi-step workflows. The first time I used this, it honestly felt like the browser was possessed (in a good way)! For example, I asked Atlas to check my webmail for any urgent client messages. It actually logged in (with my prior authorization), skimmed my inbox, and drafted a couple of email replies for me to approve. OpenAI has essentially given ChatGPT a “mouse and keyboard” inside Atlas. Now, there are safeties – Atlas will always ask your permission before, say, completing a purchase or logging into a new site. And you can run it in a restricted mode where it won’t use any of your saved logins if you prefer. But with permission, it can do almost anything you could do in a browser. In a live test, I had Atlas find the top-rated noise-cancelling headphones under $200 and attempt to buy them. It searched multiple sites in parallel, compared reviews and prices, chose a pair on sale, added it to a Best Buy cart, and got all the way to the checkout page – pausing to let me confirm the final purchase. This took maybe a minute or two, versus the 30+ minutes I might spend researching manually. It wasn’t flawless – at first it told me the headphones weren’t on Amazon (they were, it just clicked a bad link), so there is still a need to supervise and correct the AI at times. But the productivity gain is obvious.

• “Browser Memories”: Atlas introduces a feature called browser memories that personalizes the experience. Essentially, Atlas can remember what you’ve done in past browsing sessions (which sites you visited, what you asked the AI, etc.) and use that to improve future answers. For instance, after a day of using Atlas, it “learned” that I often check a specific project management dashboard and my Creatuity email each morning. The next day, it proactively suggested with a single click it could open my email and give me a summary of overnight messages. It’s optional – you can turn off memory or use an incognito mode if you don’t want the AI to learn from a session. By default, Atlas also does not use your browsing data to train OpenAI’s models (a big privacy point they stress). But if you allow it, the personalization can be incredibly convenient. It’s like your browser starts anticipating your needs, not in a creepy ad-tracking way, but in a “we remember what’s useful to you” way.

OpenAI’s CEO has described Atlas as the first major browser innovation in a decade, and indeed it feels like a big leapwired.comwired.com. One more thing to note: Atlas is currently available for macOS (Windows and mobile are in the works), and it’s free to use. The basic ChatGPT sidebar works for everyone; the fancy Agent Mode is what requires a paid ChatGPT plan. Given OpenAI’s resources and the ubiquity of ChatGPT, Atlas is definitely the one to watch. It essentially combines the strengths of Comet and Dia: it can act autonomously to perform tasks, and it provides on-demand assistance and answers as you browse.

Other AI-Infused Browsers to Watch

Beyond the big three above, a mini “browser war” is unfolding, and a few other players deserve mention:

• Opera Neon (AI Edition): Opera (yes, that browser you might remember from years ago) has been experimenting with a concept browser called Neon. In 2025, they infused Neon with an AI co-pilot. Opera’s goal is to have the browser proactively assist users and even take actions when asked – similar in spirit to what Atlas doesibm.comibm.com. Opera has a much smaller market share, but they’ve often innovated in browsing (they pioneered things like tabs and pop-up blocking back in the day), so Neon is an interesting playground for AI ideas.

• Microsoft Edge: Edge has had the Bing Chat AI in a sidebar since early 2023, making Microsoft one of the first movers to add conversational search to a browserwired.com. More recently, Edge can try to summarize pages or rewrite text you highlight, thanks to Bing AI. However, Edge’s AI is more question-and-answer and doesn’t really act autonomously in the way Comet or Atlas do. Microsoft is likely to keep integrating OpenAI’s tech (since they invest heavily in OpenAI), so I expect Edge to quietly gain more AI prowess.

• Brave: The privacy-focused Brave browser added an AI summarizer for webpages and also conducted some research into AI-driven security. Brave hasn’t gone as far as to add a full assistant, but they are integrating AI in niche ways (e.g., summarize this article in one click). It shows that even privacy-centric products see some demand for on-page AI helpGoogle Drive.

• Google Chrome: This is the big elephant in the room. Google isn’t sitting still. They’re testing an AI feature called “Search Generative Experience” in Chrome, and they announced a forthcoming Gemini AI chatbot that will live right in Chrome’s toolbar as a sparkly icon. In typical Google fashion, their initial approach is cautious – adding AI on top of search – rather than re-inventing the whole browser UI. But given Chrome’s dominance, if Google flips the switch on deeper AI integration (and they will, to protect search revenue if nothing else), that could rapidly bring AI browsing to the masseswired.com. Keep an eye on Chrome updates over the next year.

The takeaway here is that the “browsing layer” is the new competitive front in tech. It’s not just about search engines anymore; it’s about who controls the experience where users start and accomplish their tasks. AI has opened a crack in the dominance of Chrome/Safari – and everyone from startups to enterprise giants (did I mention Atlassian paid $610M for a browser?) is trying to seize this chance. For e-commerce professionals, this means our playbook for attracting and serving customers online is about to get some major rewrites.

AI Browsers in Action: How They Handle Shopping (Live Examples)

Talking about these capabilities in abstract is one thing, but seeing is believing. I want to share a couple of real examples I tried, which illustrate how an AI browser changes the shopping experience – for better and, occasionally, for worse. These examples were part of a live demo on our podcast, and they opened my eyes to just how much the buyer’s journey can be compressed.

Example 1: Finding the Best Headphones Under $200

Scenario: I asked each AI browser a simple consumer task: “Find the best noise-cancelling headphones under $200 and buy it.” In a normal world, I’d have to search that query, click maybe a Wirecutter or CNET article, read some reviews, then find a retailer, etc. Let’s see what happened with AI:

• OpenAI Atlas: Upon hearing my request, Atlas’s ChatGPT agent sprang into action. It simultaneously opened several tabs – I saw it pulling up a few “best headphones under $200” review articles and searching Amazon and Best Buy. In its sidebar, it was “thinking” through the options. Within about a minute, Atlas responded (in the chat) that the Anker Soundcore Space Q45 headphones were a top choice based on reviews, and that they were currently on sale for $99 at Best Buy (down from $149). It had already added the item to my Best Buy cart and asked, “Would you like me to proceed to checkout?” I was floored. It found a highly-rated product and the best price (even price-checking Amazon, which was higher at that moment) and got me right to the brink of purchase. Now, I did ask a follow-up: “Are these available direct from Anker or on Amazon? Is Best Buy really the best total price?” The AI quickly searched the Anker (Soundcore) official site and confirmed the price there was $149 (with a coupon) – so still higher than Best Buy – and that Amazon’s price was currently higher too. In this case, Atlas saved me money by steering me to Best Buy’s sale. I did catch a hiccup: initially the AI had said the item wasn’t available on Amazon, which wasn’t true; it turned out Amazon had it, but perhaps a broken link or a marketplace seller confused the AI. When I pressed, it corrected itself and even showed me the Amazon listing (which was $129 at the time). This highlights an important point: AI browsers can make mistakes or miss information (just like a human might), so there’s still a need for the user to stay alert, especially for critical decisions. But overall, from a single sentence request to “item in cart” – that’s a radically streamlined funnel. One continuous conversation replaced what would normally be a multi-step, multi-site journey.

• Perplexity Comet: Comet tackled the same task in a very similar manner. It, too, ended up recommending the Anker Soundcore Q45 and navigated to Best Buy’s site. However, Comet stopped short of completing the purchase on its own – it told me the item was in the cart but that I’d need to log in or checkout as a guest myself (Comet has been explicitly designed to be cautious about final purchase steps unless it can use a streamlined method). In terms of speed, Comet was extremely fast at collating information – it scanned about 20 different sources/reviews almost instantly to make its recommendation. The user experience, though, was a bit less transparent than Atlas; I mostly interacted through the chat, and it gave me the summary answer with less of a “show your work” feeling. Still, Comet proved it could handle the research and selection perfectly well. It identified the same product choice, which tells me that many AI systems are converging on using similar data (perhaps all these models ingested the same review articles or consumer ratings to decide what’s “best”).

• Arc’s Dia: Dia approached this scenario differently because it isn’t built to fully automate a multi-site workflow in one go. When I gave Dia the query, it quickly returned a recommendation (it mentioned a model of Sony headphones around $199 as a top pick, along with the Anker Q45 as another option). Instead of adding to cart, it simply provided me with links – one to the Amazon page for that Sony pair, and one to Best Buy for the Anker, etc., along with a brief on why each is recommended. Think of Dia’s style as “AI-assisted comparison shopping.” It didn’t do the cart/checkout for me, but it gave me the key info to make a choice without me having to trawl Google. And Dia was fast – faster than the others – likely because it wasn’t going as deep into the process. One neat thing: because Dia’s AI lives in the browser, if I clicked the Amazon link for the Sony headphones, I could then ask Dia follow-ups right on that Amazon page (like “Is this a good price or do prices drop often?”). In effect, Dia can accompany the user through a traditional click-based journey, making that journey smarter at each step.

Takeaways from Example 1: All three AI browsers drastically shortened the path from question to decision. In two cases, the AI even went through the motions of purchase on my behalf. This is both exciting and a bit scary for e-commerce businesses. It’s exciting because if your product is truly the best match, the AI can surface it and even drive the sale with minimal friction – potentially increasing conversion rates for those recommendations. It’s scary because if you’re not in that AI’s shortlist, the user might never see you at all. Also, brand loyalty can take a backseat: I as the user didn’t specify any particular store, I just wanted the “best place” to buy. The AI decided Best Buy was that place due to price. So in this instance, Best Buy got the sale, and Amazon lost out – despite me typically being an Amazon-centric shopper. Imagine millions of these AI-driven choices happening; the dynamics of competition could shift quickly (more on this later). One more observation: all the AIs double-checked with me before finalizing a purchase (and Atlas even highlighted it won’t input sensitive info without permission). That’s comforting from a user trust perspective. We’re not at the stage of completely hands-off purchasing – and maybe users wouldn’t want that anyway for most things. But we are at the stage of “one conversation = one converted order.” That’s a big change from the multi-click conversion funnel we’re used to measuring.

Example 2: A B2B Purchasing Task – Sourcing Construction Materials

I also wanted to see how AI browsers handle more complex, B2B-oriented queries. For instance, if I’m a construction project manager looking to source materials quickly, can an AI browser help? I tried this scenario: “I need one ton of #3 rebar delivered to a job site in Plano, Texas next week. Find the best place I can buy it online today.” This is a tall order because B2B queries often involve specialty suppliers, maybe no straightforward e-commerce checkout, etc.

Atlas took the lead on this one in Agent Mode. It started searching local suppliers, looking at construction supply companies and even Home Depot/Lowe’s. It found a few promising leads: one local supplier in the Dallas area that sells rebar (though their prices weren’t listed online), another that had a quote request form, and an online supplier that listed a price for a ton of rebar but was based in California. Atlas summarized three options with pros/cons: Supplier A is 15 miles away, likely can deliver within a week, but you have to call for a price. Supplier B offers online ordering but is out-of-state (price $X plus unknown freight, might not meet one-week timeline). Supplier C can drop-ship with quick delivery if you fill out a form. This was impressive – it gathered information I’d normally have to call around or dig through forums to get.

I then nudged the AI: “I don’t have time to call anyone; I need to order it online with pricing visible.” This made it filter differently. It actually ended up finding a metals e-commerce site that would let you add a large quantity of rebar to cart and get a freight shipping quote. The AI tried to navigate that, but it struggled a bit with the freight calculator (understandable, those can be tricky even for a human). Ultimately, Atlas suggested an online metals retailer that had the product listing and noted I’d have to await a freight quote after checkout. Not a clean one-click solution, but it gave me a path.

Lessons from Example 2: B2B e-commerce still has some catching up to do to be AI-friendly. The AI could parse and present options, but it couldn’t fully complete an automated purchase because so much in B2B happens via quotes and custom logistics. However, from a buyer’s perspective, the AI saved me a ton of time by aggregating the research. I didn’t know those suppliers offhand; the AI found them. For B2B merchants, this is both a challenge and an opportunity. Challenge: if your information (pricing, delivery capabilities) isn’t online or is buried, an AI agent might skip over you or simply tell the user “you’ll have to call them” (which many users won’t bother to do). Opportunity: if you do offer transparent online ordering or fast quote generation, you could become the go-to recommendation that the AI gives to busy buyers. In my test, the AI clearly favored the source with an online price (even though it was far away) because that data was accessible. This hints that data availability and ease of access are going to be key in winning AI-driven business.

After running these demos, I can confidently say AI browsers aren’t just hype – they really work, and they’re here now. Yes, they occasionally fumble, but so do human users. The speed and convenience are undeniable. From an e-commerce lens, I saw how an AI could collapse the classic awareness → consideration → purchase funnel into basically one or two prompts. That’s mind-blowing. It means we have to rethink how we attract the AI’s “attention” in that first prompt and how we ensure the AI can confidently choose our business to fulfill the user’s need.

Now, let’s pivot from these examples to broader implications. What do AI browsers mean for the way customers find products, visit websites, and ultimately make purchases?

Implications for E-Commerce Discovery and User Experience

If you take away one thing from this article, let it be this: AI browsers are changing where and how customers discover products and information. The power is shifting from traditional search engines and website browsing to AI-driven conversations. Here are some of the biggest implications I see for e-commerce teams:

1. “Zero-Click” Information – Answers Without Visits

When an AI browser can answer a user’s question directly, the user might never click through to a website. We already saw this trend with Google’s featured snippets and answer boxes, but AI takes it to a new level. Imagine someone asks, “What are the top 5 laptops under $1000?” An AI browser might just list them out, with brief summaries, all in the chat itself. The user didn’t have to click a “Top 5 laptops” blog or a review site – the AI already did that reading for them.

For site owners, this could mean less traffic and fewer site visits from informational searches. Your content could be getting read and used by the AI (which is good, it means you’re relevant), but you may not see those users on your pages. This zero-click behavior has been observed with AI search tools already – one report noted that click-through rates are significantly lower when people use AI search, because the AI often satisfies the query on the spotibm.com. So, part of your audience might get what they need from the AI summary and move straight to a purchase or decision without ever gracing your beautifully designed homepage.

However, I’ll add a nuance: if the AI’s answer leads into a transaction, then traffic might decline but conversion rate for that traffic could shoot up. In the headphone example, the AI dropped me off on Best Buy’s site with an item already in my cart. By the time I land there, I’m highly likely to complete the purchase – much more likely than a casual browser who came via a generic Google search. So we may see a scenario where overall site visits drop, but the quality of visits (in terms of purchase intent) goes up. E-commerce analytics teams will need to adjust their benchmarks and look more at outcomes than raw sessions, because an AI-curated session is a very different animal from a traditional organic session.

2. Discovery is Now a Conversation (and If You’re Not in it, You’re Invisible)

Traditionally, discovery meant SEO (ensuring you rank in Google for relevant keywords) or paid ads, and maybe social media buzz. In an AI browser world, discovery is happening through direct Q&A conversations. The user might not be typing the exact keywords that match your product pages; they might be asking a natural language question like, “What’s a good gift for a 5-year-old who loves science?” The AI will then compile a few product suggestions to answer that. If your product or brand isn’t among the suggestions the AI provides, that user will likely never know about you. There’s no second page of results, no chance for them to browse further down – it’s one and done, in many cases.

This raises the stakes for what we could call AI-era visibility. Just as brands have fought to be in the Amazon “Buy Box” or in the top Google results, we will need to fight to be in the AI’s shortlist of answers. The ranking factors for that are still evolving, but we have some hints. OpenAI has shared that for its ChatGPT shopping feature (which Atlas uses), it ranks results by relevance and also by factors like availability, price, quality, whether the seller is the primary (official) seller, and whether instant purchase is enabledopenai.com. In other words, if two sites both sell the same widget, the one with the item in stock, at a lower price, with better reviews, and the one that supports frictionless checkout (like ChatGPT’s Instant Checkout), will be favored. That’s a new SEO of sorts: call it “AI SEO.”

Also, structured data is becoming more crucial. AI models don’t literally scroll and read the web the same way a human does. They often rely on structured data feeds (like product feeds, schema markup on pages, etc.) to quickly ingest information. If your site has excellent schema.org markup – say your product pages clearly tag the name, price, description, review scores, etc. in JSON-LD format – an AI can easily pick that info to answer a question. If your site lacks that, the AI might skip over you in favor of a competitor whose data is easier to parse. We’ll talk action steps soon, but I strongly suspect structured data and real-time inventory feeds are about to become as important as traditional SEO keywords were in the past. They are your ticket to being included in AI answers.

3. Compressed Funnel, Changed UX

Think about the classic e-commerce marketing funnel: Awareness → Consideration → Decision → Action. AI browsers compress a lot of that into a single conversational flow. A user can go from not knowing what they want, to getting a personalized recommendation, to purchasing – all without leaving a chat window. That means the role of your website’s design and content may shift. The “landing page” of the future might not be a webpage at all, but rather a snippet of content or data that an AI pulls from your site to present to the user.

For instance, instead of someone landing on a product detail page and reading it, the AI might just say, “I recommend the Acme SuperWidget 3000, it’s $249, has a 4.8-star rating from 500 reviews, and ships within 24 hours.” The user might then ask, “Is it compatible with iPhone?” and the AI might say “Yes, according to the manufacturer’s FAQ.” The user then says “Okay, buy it.” In that whole sequence, the user didn’t visually see the Acme product page or the FAQ page; the AI retrieved and relayed the info. From a UX perspective, the AI was the interface, not your website.

This means we have to optimize our content for AI parsing. Think about how your product information is presented: is it locked up in long paragraphs and marketing-speak? Or do you have a clear Q&A, bullet points, spec tables, and schema markup that an AI can easily digest? I’d wager that sites will start adding an “AI summary” or “AI metadata” section – not visible to humans, but structured for AI – so that when an agent visits, it gets exactly the key points and nothing else. In fact, some retailers are reportedly experimenting with AI-specific landing pages already: pages that look ugly or minimal to humans but are perfect for an AI to parseGoogle Drive. It sounds extreme, but if an increasing percentage of your “visitors” are AI bots fetching info for users, it might make sense to feed them what they need separately from your human-centric pages.

Another UX change: branding and storytelling may get intermediated. If I’m interacting only with ChatGPT or Comet, I’m not necessarily seeing the beautiful images on your site, the branding, the tone of voice you crafted – I’m getting a distilled data-driven answer. This is a bit of a double-edged sword: it levels the playing field for smaller brands (because the AI might surface the objectively “best” product even if the brand is not well-known or isn’t paying for placement), but it also risks commoditizing the experience. As a brand, you have to work harder to ensure your value propositions and identity come through in the data that the AI sees. For example, if you pride yourself on eco-friendly materials or lifetime warranty – make sure that’s in the product description or schema markup in a way that the AI might incorporate into an answer (e.g., “this jacket is environmentally friendly and comes with a lifetime guarantee”). Otherwise, the AI might just say “here’s a jacket for $99” with no sense of your brand differentiators.

4. Rethinking Advertising and Marketing

A big question: How will digital advertising work when the AI is the one choosing what to show the user? Right now, companies pay for Google Ads to be the top result or for Facebook ads to appear in feeds. If users are asking an AI browser for recommendations, you can’t exactly buy “ad space” in the traditional way. The AI isn’t going to blatantly say “here’s a sponsored result” unless that becomes an accepted norm (which could happen, but it’s tricky – if the AI’s primary job is to “help,” injecting ads could degrade trust).

One possibility is AI browsers might introduce sponsorship or preferred placement in subtle ways. Perhaps if a user’s query has multiple good answers, the AI might add, “(and by the way, [Brand] has a special offer on their top-rated model right now)” if [Brand] paid for that privilege. But this is speculative. In the immediate term, the focus for marketers should be earning organic recommendations from AIs. That means feeding the AI the data that makes your product a logical choice (best price, great reviews, etc., as discussed). It’s akin to optimizing for Amazon’s algorithm – you want to be the recommended product, not necessarily by paying, but by meeting the criteria the algorithm values. In a way, it’s a return to true merit-based marketing: be the best option for the customer and the AI will pick you. Of course, how the AI determines “best” is the secret sauce we’re all trying to figure out. But from OpenAI’s disclosures, it’s a mix of objective factors like price/availability and qualitative factors like product quality and seller reliabilityopenai.com.

One interesting development: ChatGPT now has a feature called “Shopping Demo (Beta)” or sometimes called ChatGPT “Product Finder”, which integrates with certain merchants (like Shopify stores and others via the Agentic Commerce Protocol). I’ve been using a preview called ChatGPT “Pulse”, which actually pushes product recommendations to me unprompted. For example, I mentioned earlier that I had been talking with ChatGPT about photography. One morning, my ChatGPT Pulse proactively suggested: “Hey, the newer model of your camera just went on a big sale, here’s the price and a link if you’re interested.” That was not me searching – that was the AI recommending a purchase based on what it knows about me (my camera is older, I love photography, and a deal emerged). That’s a new kind of “advertising” – more like personal concierge service. It wasn’t paid placement; it was an organic recommendation (though one could imagine future deals where brands feed their promotions into such systems). The key point: AI might shift us from a pull model (user searches for something) to a push model (AI suggests something before you even ask). For e-commerce, that opens both opportunity (catch the customer at the perfect moment) and complexity (you need to have your data and integration set so that AI knows about your promotions or new products).

Finally, consider the design of your site in an AI world: If most human visitors coming via AI are essentially at the bottom of the funnel already (e.g., coming to your site just to hit “Buy” or to maybe see one detail the AI pointed them to), do you need to simplify those experiences? Maybe the AI will send users deep-linking into a checkout page or a specific FAQ anchor. Your site’s job might become less about persuading and more about seamless transaction processing and trust signals (because presumably the AI already did the persuading). Also, performance and accuracy are paramount – if your site is slow or returns incorrect info via an API, the AI might drop you. As one expert bluntly put it, “It changes SEO forever… Now we’re moving into AI SEO. How is that landing page even relevant now? Is there a landing page that is designed for Comet browsing?”ibm.com. In other words, we have to start designing experiences not just for human eyes and clicks, but for AI agents that read and interact with our sites.

Six Action Items for E-Commerce Directors (Start Now)

Alright, this is a lot to take in. If you’re an e-commerce director or tech decision-maker, you might be thinking: “This is fascinating (or alarming), but what do I do about it?” Here are six practical steps I recommend you take in the next few weeks to prepare for the AI browser era. Whether you have a big budget or none at all, these are actions that mostly involve time and strategy, not necessarily huge spend. They will get you moving in the right direction:

1. Get Hands-On with AI Browsers – There’s no substitute for firsthand experience. Download and try out OpenAI Atlas (if you have a Mac, as of now) or Perplexity Comet. If you use Windows, give Microsoft Edge’s Bing Chat a spin; if you can get an invite to Arc/Dia, try that too. Use them as if you were a customer. For example, search for your own brand or product in these AI tools. Ask, “What’s the best [your product category]?” or “Where can I buy [your brand name] [product]?” See if and how the AI mentions you. This can be eye-opening. If the AI has never heard of your brand or has incorrect info, that’s insight you wouldn’t get without testing. Also, try a few of your common customer queries and see what answers come up – are they citing a competitor’s blog or a random site? This will highlight content gaps or reputation issues you may need to address. The bottom line is to familiarize your team with how these AI browsers operate. It’s one thing to read about it, but when you use it, you start thinking of all sorts of new ideas and concerns specific to your business. (On a personal note: once I got used to Atlas, I found myself instinctively wanting to ask my browser to do things even when I was back in Chrome. That’s how quickly it can shift your mindset.)

2. Structure Your Data for AI – This is a foundational step. Ensure your website has up-to-date, comprehensive structured data for all your products and content. That means implementing schema.org markup (typically via JSON-LD scripts in your page HTML). At minimum, use Product schema on product pages (including name, description, price, availability, SKU, reviews, etc.), FAQ schema on Q&A pages, Article/Blog schema on content pieces, and so on. Why? Because structured data is like speaking in a language that AI agents natively understand. It builds trust – the AI can easily verify facts about your products from your own site if you present them clearly. If, say, your page says “Free Shipping” in human-readable text but you also have a shippingPolicy schema that indicates free shipping, an AI agent can confidently tell a user “this item has free shipping” without misinterpreting. Likewise for stock levels, pricing, etc. Many e-commerce platforms have plugins or built-in features for schema – use them. Also, if you have an internal search or product API that’s accessible, consider exposing a feed or endpoint that AI platforms (or even just Google’s crawler) can use for up-to-date data. Trust between AI and your site starts with structured data – if the AI can’t easily figure out what you sell, at what price, and how well it’s rated, it will look elsewhere.

3. Feed the AI with Your Product Data – Beyond on-page schema, look into providing direct product feeds to AI services. OpenAI has introduced a system where merchants can integrate their product catalogs into ChatGPT’s knowledge base (often referred to as the ChatGPT merchant or plugin program). Shopify, notably, has already partnered such that Shopify’s millions of merchants will have their products discoverable by ChatGPT by defaultopenai.comopenai.com. If you’re on Shopify, ensure your listings are updated – you might already be part of this without any extra work. If you’re on another platform (Magento/Adobe Commerce, BigCommerce, etc.), you may need to apply to OpenAI’s program or similar and provide a product feed. This usually involves generating a JSON or XML feed of all your products with details like title, description, price, URL, images, etc. – similar to what you’d do for Google Shopping or Facebook Catalog. The Agentic Commerce Protocol (ACP) introduced by OpenAI is an open standard for hooking into ChatGPT’s instant checkout systemopenai.com. Even if you’re not ready to implement full “chat checkout,” getting your products indexed is the first step. The message here is: don’t wait for AI to hopefully stumble on your site; proactively push your catalog to the AI platforms. Keep those feeds fresh (update them whenever prices or stock change). Some early-adopting merchants are finding that having a direct feed means ChatGPT will recommend their products more accurately and frequently, because it knows it can trust that data. On the flip side, if your data is inaccurate or stale, the AI might present wrong info or lose trust in your brand, making it less likely to show you next timemagebit.com.

4. Optimize for “AI Ranking” Factors – We touched on this, but it’s worth reiterating with action items. Make sure the attributes that an AI would consider “signals of a good result” are front and center in your data. Some things to focus on:

   • Competitive Pricing: If you can’t always be the lowest, consider at least having price-match guarantees or highlighting quality differences. AI will likely weigh price heavily (because users ask for “best” which often implies value). If you are the premium option, have data to justify it (e.g., higher review score, better materials – ensure those points are known to the AI).

   • Availability: Nothing will get you booted from an AI recommendation faster than being out-of-stock. If your feed or schema says out of stock, the AI might skip you. Keep inventory status updated. If you have backorder or quick restock, indicate expected ship times so the AI might still consider you (“usually ships in 3 days” is better than a flat “out of stock” with no info).

   • Shipping & Fulfillment: Fast and free shipping are likely plus factors. If you offer free shipping or free returns, put that in your metadata (e.g., schema hasMerchantReturnPolicy or simply in the description text the AI might read). ChatGPT’s system even considers if you as a merchant are known for reliability and fast fulfillmentopenai.com. While there’s no public “reliability score”, you can bet that if an AI can choose between a merchant with 95% positive reviews and on-time delivery vs another with 80%, it’ll favor the former. So investing in customer satisfaction and encouraging reviews feeds into this.

   • Reviews and Ratings: Speaking of reviews – ensure your product ratings are aggregated and visible. Use schema for aggregateRating. If you have great reviews, the AI will often quote them (e.g., “4.8 stars from 300 customers”). That’s a selling point the AI will use on your behalf. If your reviews are weak, well, that might hurt you in AI ranking just as it would in human decision-making.

   • Instant Checkout Enabled: This one is new – OpenAI indicated that if a merchant supports its Instant Checkout (via the Agentic Commerce Protocol), it can use that as a factor in ranking (not to bias the result, they claim, but to “optimize user experience”)openai.com. In plain terms: a merchant that lets ChatGPT complete the purchase in-chat provides a smoother UX, so ChatGPT might prefer to show that merchant over another, all else equal. If you have the means, consider implementing these emerging standards (like ACP). Even if you don’t yet, be aware that the convenience factor matters.

   • Primary Seller Status: If you’re a brand that sells direct, emphasize that (e.g., “Official Store” in your feed or site name). AI might give weight to the “primary” seller or manufacturer site versus a third-party reselleropenai.com, under the assumption that the primary source has the latest info or better support. On marketplaces, ensure your brand is clearly indicated so the AI knows it’s an official listing.

5. Rethink Your SEO and Content Strategy for AI – Traditional SEO (targeting specific keywords, building backlinks, etc.) isn’t going away overnight, but we need to expand our content strategy to speak the language of questions and answers. Start creating content (or reformat existing content) that directly answers the kinds of natural language questions your customers might ask an AI. Some tactics:

   • FAQ Pages: Build out robust FAQs for your products and your domain. If you sell outdoor gear, have an FAQ like “How do I choose the right hiking boots?” where you, as the expert, provide a concise helpful answer. These are the snippets an AI might pull when a user asks a similar question. If the AI ends up quoting your site as the source of truth, that’s a big win (and might still drive the user to click through for more).

   • How-To and Guide Content: Write guides in a Q&A style. For example, a title like “Top 5 Questions About Choosing a Smartphone in 2025 – Answered” could be gold for AI. The AI might use those clearly labeled Q&A pairs to answer users. Also, content that is conversational and factual (rather than just marketing fluff) will be favored by AI.

   • Be the Source to Cite: Today, some AI like Bing will cite sources with links. Even ChatGPT’s browsing mode will list sources it used. If your content is high-quality and factual, you increase the odds that the AI will cite you as evidence. That’s a new kind of SEO—AIS (AI Source) Optimization, if you will. It means writing content that’s not just aimed at ranking, but at being trustworthy and quotable.

   • Monitor AI Queries: This is tricky since we don’t have “AI Search Console” yet, but pay attention to what questions customers are asking your support teams or chatbots. Those are likely the same questions they might ask AI. Also, some SEO tools now attempt to show popular questions or AI-generated queries. Use that intel to craft content. For instance, if a lot of people ask “Is brand X compatible with brand Y?”, and you sell those, make sure that info is clearly answered on your site.

   • Don’t Neglect Traditional SEO Completely: We’re in a transition. Many users will still come via Google for now, and Google’s own AI summaries often draw from the top SEO results. So continue good practices like page speed, mobile optimization, and earning reputable backlinks – those help both old-school SEO and indirectly your AI visibility (since AI often trusts the same sites that have good SEO presence and authority).

6. Monitor, Learn, and Adapt – Finally, treat this as an ongoing priority, not a one-and-done project. Assign someone on your team (or a few folks across teams) to specifically monitor AI browser trends and their impact on your metrics. Things to do:

   • Analytics Watch: Keep an eye on organic traffic and referral patterns. You might notice changes, like fewer clicks from search engines but maybe more direct or untracked referrals (some AI agent traffic might appear as direct or weird referral sources). If you see conversion rates climbing while sessions drop, that could be an AI effect (fewer people physically coming, but those who do are highly intent because an AI sent them ready to buy).

   • Customer Feedback: Add a question in post-purchase surveys or feedback forms: “How did you hear about us?” or even specifically, “Did you use any AI assistants or ChatGPT to help you make your decision?” It sounds futuristic, but you may start seeing answers like “I asked ChatGPT” or “Found via AI browser.” This qualitative data is gold. It tells you if all this really is affecting your customers or if it’s still niche. Early detection of an “I used AI” trend among your buyers can justify investing more in those channels.

   • Keep Up with AI News: Subscribing to newsletters or podcasts (like Commerce Today 😉) that track AI in commerce can help you stay informed. The landscape is evolving weekly. For example, a month ago we didn’t know Atlas was launching; now we’re all over it. Next month, there could be a new GPT model or a new partnership (what if Amazon launches their own AI browser, or Adobe embeds an AI shopping assistant in PDF catalogs – who knows!). Having a pulse on the developments means you won’t be caught off-guard.

   • Experiment and Iterate: Consider running some experiments. For example, update schema or content on a subset of products to be more AI-friendly, then see if those products start getting more mention or sales via AI referrals. If you have the ability, maybe create a basic “AI landing page” hidden on your site that’s richly structured and see if an AI agent finds it useful. Share findings with your team and iterate. We’re all learning here – there’s no playbook yet, which is both exciting and challenging. The winners will be those who learn the fastest.

Closing Thoughts: Designing for a World of AI Browsers

AI browsers aren’t a sci-fi concept or a prototype in a lab – they’re here now, and they’re already changing user behavior in commerce. From what I’ve observed, early adopters (especially the younger, tech-savvy consumers and busy professionals) are embracing these tools to save time and effort. Mid-market e-commerce teams – which often have the advantage of agility – can get ahead by experimenting early and adapting. This is a classic case of disruption where being proactive now could translate into outsized gains in customer acquisition and loyalty down the line.

The key mindset shift is to start designing your digital presence for two audiences: humans and AI intermediaries. In the past, we optimized for humans and then maybe for Google’s crawler as a secondary consideration. Now, the AI agent might often come before the human or act as a gatekeeper to the human. If you cater well to the AI (with accurate data, quick responses, integration hooks), you ultimately cater to the human by proxy because the AI will serve them better on your behalf. Think of the AI as a very important new customer – one that looks at your site in a very different way than a person does. As Chris Hay of IBM astutely asked, “Is there a landing page that is designed for Comet browsing?”ibm.com. It’s a provocative question. We might soon see websites with sections or pages tailored only for AI consumption, whose sole job is to ensure the AI has what it needs to represent the brand well. Perhaps that’s what comes next in web design.

In closing, I encourage you to lean into this change. Try these browsers yourself, show your team, maybe even do a lunch-and-learn demo. Spark discussions about how your e-commerce operations can leverage AI and how to mitigate the risks. The companies that adapt will provide smoother, smarter experiences for customers and will thrive; those that dismiss it (“oh, users will never change their habits”) may wake up to find their hard-won SEO ranking or brand recognition isn’t translating to sales like it used to.

If you’re feeling overwhelmed or unsure where to start, remember you’re not alone – and you don’t have to navigate it alone either. At Creatuity, we’ve been diving deep into AI and commerce, and we’re helping clients chart their paths in this new landscape. I’m personally immersed in this topic daily (as you can probably tell from this lengthy post!). If you want to discuss how AI browsers and ChatGPT shopping could impact your specific business – or if you need help implementing some of the action items I listed – feel free to reach out. I’d be happy to brainstorm with you. Just hit the “Schedule” button at the top of our website to book a consultation with me and my team. Let’s embrace this new era of e-commerce together and make sure your brand is front and center in the AI-driven future of shopping.

Thank you for reading, and I look forward to seeing how we all innovate in the age of AI browsers!