Business of Fashion runs an opinion asking whether AI will be online shopping's next victim. This is entirely the wrong frame. AI won't kill online shopping — it will kill the part that deserves to die: the fifteen-tab comparison session, the keyword search returning 4,000 results, the scroll-forever-and-give-up product page that is still the majority of e-commerce in 2026. What survives is either the AI front-end (discover, decide, buy, without ever visiting a product page) or the experience front-end (tactile, event-driven, human-assisted, genuinely irreplaceable by software). The thin middling layer — adequate search, generic navigation, no real service — is what's actually endangered.

This is the same squeeze Shopify's platform AI layer is already applying from below, and it's the fault line David's Bridal and The Knot are navigating from opposite ends of the bridal aisle. The question is not whether AI kills shopping. The question is which layer of the retail stack survives when the friction it existed to manage disappears.

New e-commerce trends data says 60 percent of online shoppers want AI-powered tools. Read alongside the BoF opinion above, this is a clarification, not a contradiction: consumers want AI assistance in shopping; they do not want the shopping experience replaced entirely. The gap between those two things is enormous. An agent that recommends your size is welcome. An agent that selects your outfit and charges your card without being asked is a different product — and most shoppers aren't there yet, no matter what the agentic commerce evangelists say.

Michael Kors has added an AI-powered assistant to its website. The announcement is brief, the implementation details absent from the coverage — which is precisely the problem. When any mid-market brand can now bolt a chatbot onto a product page, the differentiator stops being the technology and starts being what the assistant actually knows. Does it know your purchase history? Does it distinguish between a loyal customer and a coupon-code arrival? Does it upsell intelligently or desperately? The interface is commodity. The training data, the intent design, the escalation logic — that's everything, and it's the part nobody mentions in these announcements.

Thorne's CSO called AI wellness chatbots "table stakes" last week. Michael Kors is arriving at the same plateau. The press release is the beginning of the problem, not the end of it.

Etailment, Germany's e-commerce trade press, runs a piece asking how retailers should prepare for the KI-Wende — the AI inflection point. The practical checklist is familiar: structured data, transparent pricing, deep API accessibility, machine-readable catalog attributes. The uncomfortable implication is that most merchants — German, European, American, all of them — are not ready on any of these dimensions.

The agent doesn't browse. It queries. If your catalog isn't machine-readable, you don't exist in agentic commerce, full stop. Honor's MagicOS 9.0 is already executing cross-app purchases automatically, meaning Chinese consumers are living in the agentic future while European retailers are still writing preparedness checklists. Shopify's Catalog API and Agentic Storefronts are the closest thing to an on-ramp for English-language merchants who want to exist when an AI is doing the choosing. The German mid-market has no equivalent. Neither does most of American retail.

Topshop staged an AI-driven shoppable catwalk, with Shark Beauty and Lookfantastic providing live TikTok commerce on top. Be precise about what this actually is: a brand that no longer owns a single store — ASOS acquired Topshop in 2021, closed everything physical — used AI visual production to stage a catwalk event, then attached live social selling to monetize the stream. This is not fashion. It's a content format with a checkout button.

The architecture descends directly from the Taobao and Pinduoduo live commerce models, transplanted into British aesthetics via a platform the original Chinese operators don't control. Pinduoduo is running AI-generated interactive dramas that route viewers to product pages. Topshop is running AI-generated runway looks that route viewers to beauty products. The underlying mechanic is identical — the west just took five years to discover it. Today's Debenhams piece covers the marketplace variant of the same logic: dead brands finding new infrastructure to haunt.

McKinsey drops a live commerce primer framed as transformation in progress. TikTok Shop already proved you can sell fragrance — the most sensory-dependent category in retail — through a screen. The Topshop catwalk above is the next evolution. Consultants arrive, as always, just after the early movers have made all the interesting mistakes and the format has already been commoditized in the markets that matter. Points for production value. Minus points for calling it a transformation when it's already infrastructure.

Southeast Asian e-commerce projected at 16% growth, driven by "video and AI" as co-factors. The framing treats them as parallel trends. They're not — the AI is what makes video commerce function at scale: real-time translation across six languages mid-stream, personalized product insertion into live feeds, automated moderation that keeps a broadcast running while the host sleeps. Shopee, TikTok Shop, and Lazada aren't running two programs. They're running one converged infrastructure that Western analysts keep describing as two separate phenomena because it's easier to file that way. Separating "video" from "AI" here is like separating "electricity" from "light."

1.5 trillion RMB is a large number that will almost certainly happen. Chinese e-commerce growth is increasingly driven by the quick-commerce and instant-delivery stack, not the classic browse-and-ship model. Alibaba took a RMB 10.4 billion EBITA hit to build that layer. Hillhouse and Temasek are holding Pinduoduo at multiples of their Alibaba positions, because the institutional money has already bet on who captures the margin inside that 1.5T. The forecast doesn't answer the question that matters. The 13F filings did.

Algolia's sixth annual e-commerce search report finds — with admirable consistency — that search is retail's top digital priority and AI investment is resilient. Algolia is a search vendor. Their survey sample is companies that agreed to participate in a search vendor's study. Take the directional signal, not the precise rankings.

The directional signal, confirmed by enough independent sources to be credible: brands are protecting AI spend even as budgets compress elsewhere. Modern Retail's research found AI and budget cuts are hitting agencies as the same line item — meaning money is moving from headcount to software. That shift is real regardless of who measured it. The question Algolia's data cannot answer is whether the AI search spend is converting. That part stays buried.

One in eighteen pounds of UK retail spend flows through Amazon's third-party sellers. This is not Amazon's revenue — it's the GMV of businesses using Amazon's rails and paying Amazon's compounding fees to do so: listing fees, fulfillment fees, advertising bids to appear in results they're already paying to occupy. The gravity is extraordinary. The 5.5% share also explains why sellers don't leave despite the economics: the alternative is rebuilding discovery from scratch. Shopify's platform AI is the most credible off-ramp, and it's still building the runway. Until discovery outside Amazon is demonstrably equivalent, the platform tax is just the cost of existence.

Zalando is launching a dedicated shopping app in the UK. An app means push notifications, location behavior, session patterns, dwell time, return habits — all of it feeding directly into the personalization stack that Zalando has spent years building and the Co-CEO has been publicly discussing as a competitive moat. The UK fashion app market is contested by ASOS, Next, M&S, and every fast-fashion import with a TikTok budget. Zalando's differentiator is supposed to be data-sophisticated curation rather than race-to-the-bottom price. This app is where it tests that claim in one of Europe's most skeptical fashion markets. The data collection starts on day one; the curation proof will take longer.

Onton, an AI-powered product discovery and recommendation platform for e-commerce, has raised $7.5 million. The round is small by AI-cycle standards, but capital is still flowing into this specific problem because the thesis hasn't changed: whoever controls what the customer sees first controls where the margin goes. After Yupp.ai's $33 million shutdown last year — dead because the market it was built for pivoted to agentic systems before the product could catch up — investors are still writing discovery checks, just at tighter valuations and narrower scope. Which either means the opportunity is genuinely durable, or the lesson from Yupp hasn't fully landed.

Reale Actives launched March 31, sold out, and received a genuinely mixed response. The sell-out was never in question — Alix Earle is essentially a human distribution algorithm, routing ten million TikTok followers toward whatever she points them at. The mixed reception is the more interesting signal: it means the brand promise was evaluated, not just purchased. Sell-out velocity answers "did people want to try it." Repurchase, which we'll know in sixty to ninety days, answers "did it work." Creator-brands have solved the launch problem almost universally. Most still haven't solved the product problem.

This is the same dynamic at Forta (Lexie Hull's performance makeup) and at It's a 10 — which today is bringing in Khloé Kardashian as the paid answer to the same distribution question. The difference is that Earle and Hull are their own channels. It's a 10 is renting one.

Marc Dambremez moves from Levi's to lead Van de Velde, the Belgian group behind Marie Jo, Prima Donna, and Rigby & Peller. Levi's brings American mass-market DNA: channel management at scale, wholesale architecture, licensing frameworks, the mechanics of running a large brand through department stores and partners you don't control. Van de Velde is a specialty lingerie group under sustained pressure from digital-native and DTC competitors who have no interest in the same wholesale channels. The hire signals the board believes the problem is distribution engineering, not product desirability — which may be correct, since Van de Velde's brands are genuinely well-regarded. Whether the Levi's playbook works in fine European lingerie will be clear within two years. The answer will hinge on whether channel restructuring is enough, or whether the entire go-to-market model needs replacing.