
From Our Sponsor:
How to onboard an AI customer service agent for your Amazon store
Every Amazon seller has the same struggle: hours lost every day answering "where's my order?" for the 400th time.
It's the highest-volume, lowest-value task in your business. And it gets worse the more you sell.
This is the exact problem AI agents were built to solve.
Not chatbots. Not auto-responders that tell your customer to "contact support" ... which is what they just did. Actual AI agents that pull the order, check your policies, and resolve the ticket. No human needed.
eDesk built exactly this. Their AI Agent works across Amazon, Shopify, TikTok Shop, and 300+ channels from one place. Leading clients already have 65% of all tickets resolved entirely by AI, and that number keeps climbing.
Every ticket it handles is time your team gets back to do work that actually grows the business.
The Click Is Being Designed Out of Commerce
Every major AI platform is quietly building a world where your website becomes optional. Here's what that means for sellers.
For most of the past two decades, the commercial internet has run on a simple deal. Platforms like Google and Meta captured attention, sold access to it, and then handed the user over to your website, where the actual selling happened. The click was the handover. An entire industry — media buyers, conversion rate optimisers, landing page designers, the people who say "above the fold" with genuine conviction — was built around that single moment.
That deal is being renegotiated. And nobody asked us.
Two developments from recent months show the same pattern from different directions. Google's AI Overviews are answering queries so completely that users have no reason to click through. OpenAI has started testing an ad format where the shopper never leaves ChatGPT at all. Taken together, they suggest the largest AI platforms are building an environment in which the user increasingly has no reason to visit your carefully optimised product page. (You know, the one you spent six weeks A/B testing the hero image on.)
OpenAI's Ad Format That Goes Nowhere

Sponsored Agents
OpenAI has begun testing something called Sponsored Agents. A user sees an ad inside ChatGPT, clicks a button labelled "Chat with us," and instead of landing on the advertiser's website, the click opens a branded AI conversation inside ChatGPT. The shopper asks questions, gets answers, considers their options — all without leaving the platform.
Their own example is a shopper looking at a dining table who wants to know whether it'll fit the room, how many people it seats, and how to look after the finish. These are exactly the questions a product detail page handles terribly and a decent salesperson handles well. The format is designed to capture the consideration stage and keep it inside the walled garden.
The test is small — a limited alpha with selected US advertisers, Wayfair among them. But the surrounding infrastructure is arriving fast. Shopify has become the first ecommerce partner for ChatGPT Ads. HubSpot is the first CRM partner. Conversion optimisation with impression-based billing is reportedly the recommended buying model. The components of a full advertising business are being assembled at a pace that suggests this is not an experiment.
Comparisons have been drawn with Meta's ads that open a WhatsApp conversation. The comparison is useful but incomplete. A WhatsApp thread usually ends with a person, or at least with the brand's own systems. A Sponsored Agent conversation happens on OpenAI's infrastructure, inside OpenAI's interface, sitting right next to OpenAI's own assistant. That's a meaningful difference, and if it's making you slightly uncomfortable, you're paying attention.
Meanwhile, Google's AI Overviews Are Eating Your Clicks
To understand why OpenAI's move matters, look at what's already happened to search.
Google's AI Overviews — the Gemini-generated summaries that now sit above organic results — have been rolling out market by market since 2024. France offered a particularly clean test case. AI Overviews launched there on 22 July 2026, and researchers already held Search Console data for close to a thousand of the country's highest-traffic domains.
The results were stark. Domains where more than a fifth of queries triggered an AI Overview saw their median click-through rate drop by roughly 23 to 25 percent. More than four in five of those highly exposed domains lost click-through rate. Across the whole panel, almost half of sites lost more than a tenth of their daily clicks. One in twenty lost half.
Two details matter especially for ecommerce. First, rankings barely moved. Pages still appeared in results as often as before. They were simply clicked less. An AI Overview doesn't remove your brand from the page; it makes your page unnecessary. That breaks a long-standing assumption in search reporting — visibility and traffic used to travel together. Impressions can now rise while clicks fall. (Your SEO dashboard has never been more misleading, and that's saying something.)
Second, the researchers found a rough rule of thumb: each additional percentage point of a domain's queries triggering an AI Overview corresponded to roughly one percentage point of lost click-through rate. It's a crude relationship drawn from nine days of data, but it gives operators an order of magnitude to plan around, which is more than most of us had last week.
Where the Protection Lies (and Why It Might Not Last)
Queries with commercial intent — where the user needs a price, a product, a purchase — have lost far less than informational ones. One aggregation puts the decline on non-branded commercial queries at around 20 percent, a fraction of the losses on informational content. Branded queries may even gain click-through rate when an AI Overview appears, though that finding rests on a narrower evidence base.
The logic is intuitive. A summary can tell someone what collagen does or how to season a cast iron pan. It can't sell them a specific tub of collagen or a specific pan. At least not yet.
And "not yet" is doing a lot of heavy lifting in that sentence. Because this is where the two stories connect and the conventional reassurance starts looking fragile.
The advice given to ecommerce brands for the past two years has been: accept the loss of informational traffic and concentrate on commercial, consideration-stage content. That's the traffic AI summaries supposedly can't absorb.
Sponsored Agents are aimed squarely at that remaining territory. That dining table example isn't informational. It's a shopper mid-purchase decision — precisely the moment that's so far justified visiting a brand's website. Google has been placing ads inside AI Mode since 2025. Amazon began letting shoppers complete purchases inside Alexa+ conversations in June. Each move takes a slightly later stage of the purchase journey and relocates it inside the platform.
The pattern, viewed in sequence, is layered. AI summaries took the informational layer. Conversational ad formats are reaching for the consideration layer. Agent-led checkout is testing the transactional layer. Your website isn't disappearing, but its job is being narrowed from "the place where a shopper learns, compares, and decides" towards "the place where an order gets processed." Which is a polite way of saying your site is becoming a checkout page with delusions of grandeur.
The Incentive That Doesn't Reverse
It's tempting to describe all this as a coordinated campaign against the open web. The more accurate description is less dramatic and more durable: every platform's commercial value depends on how much of the user's session it holds. Sending a user to your site hands that value away. Keeping them inside the interface retains it. No company needs a grand strategy to arrive at this conclusion; the incentive does the work.
That matters because incentives don't reverse the way strategies sometimes do. OpenAI's advertising business is, by most accounts, still promising rather than proven. But a format that keeps the entire conversation inside ChatGPT lets them observe and optimise more of the journey. If it works, the pressure to extend it is enormous. If it doesn't, the incentive to try something similar remains.
There's also a tension in OpenAI's positioning worth watching. They say Sponsored Agent conversations are separate from ChatGPT's own answers. The pitch to users is that the assistant's judgement isn't for sale. The pitch to advertisers is a channel sitting directly beside that judgement. Whether users perceive the distinction — and whether it holds as ad revenue grows — is one of the more consequential open questions in commerce right now.
The Bottom Line for Sellers
The click — that simple handover from platform to website — has been the commercial internet's core unit of exchange for two decades. What we're watching is its gradual elimination, not through any dramatic announcement, but through the quiet, structural logic of platforms that make more money the longer they keep users inside.
The practical takeaways are blunt. First, the ad becomes an operation, not a piece of creative. Whatever your agent can explain, compare, or get wrong is the advertisement. Brands whose product knowledge lives in scattered spreadsheets and the heads of two customer service staff aren't ready for this — the bottleneck is documentation, not technology. Second, you'll get thinner data. Inside a Sponsored Agent conversation, you see aggregated reporting and direct messages, not the full browsing behaviour you'd get on your own site. Third, your reporting needs updating. If visibility can rise while traffic falls, traffic is no longer a reliable proxy for presence. Your brand may be more visible than ever while your analytics tell you it's shrinking.
None of this is settled. Sponsored Agents are in alpha. The French data covers nine days. Shoppers might reject branded agents with the same suspicion they reserve for customer service chatbots (which, let's be honest, is considerable). But the direction is consistent across Google, OpenAI, and Amazon, and the incentives behind it are structural.
The question isn't whether the click gets designed out. It's what you own when it does.
Do You Love The AI For Ecommerce Sellers Newsletter?
You can help us!
Spread the word to your colleagues or friends who you think would benefit from our weekly insights 🙂 Simply forward this issue.
In addition, we are open to sponsorships. We have more than 70,000 subscribers with 75% of our readers based in the US. To get our rate card and more info, email us at [email protected]
The Quick Read:
OpenAI launches GPT-6 Sol and Luna, cutting API prices 50% versus their GPT-5.6 predecessors. Sol beats Claude Opus 5 on AutomationBench at 9% of the cost per task, taking the price war down the model stack.
Ahrefs data shows French sites most exposed to AI Overviews lost 23.1% of CTR in nine days after the July launch. The rule of thumb: every point of AI Overview exposure costs roughly one point of click-through rate.
OpenAI rolls out ChatGPT Ads in seven Asian markets including Indonesia, Singapore and Taiwan, with self-serve buying from day one. Ads now reach 60+ countries, but performance data still covers only the US and UK.
Madison & Wall forecasts global ad revenue hitting $1.3 trillion this year, up 11%, fuelled by AI buying tools and AI firms' own ad spend. The same boom could turn to bust as AI competes for investor capital.
TikTok Shop says livestreams drive 94% of US luxury resale revenue, with category GMV up 400% year over year. Live video does what product pages can't: builds trust in high-ticket secondhand goods.
Pinterest launches a campaign optimization score in Ads Manager, rating setups on automation, creative variety, broad targeting and steady budgets. In practice, it's a nudge to hand more control to Performance+.
The Tools List:
📜 Raycast - Mac launcher with AI, clipboard history, snippets and extensions.
🌐 Bolt - In-browser AI app builder supporting many frameworks.
📹 Crayo - Auto-generate faceless short-form videos with captions and voiceover.
🤖 Windsurf - AI-powered IDE with an agentic coding assistant.
About The Writer:

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.
For Team and Agency AI training book an intro call here.




