Non-English scoop, and today's best one. etailment.de is asking the operational question the English-language agentic commerce keynote circuit keeps skipping: when an AI agent completes a checkout — user-authorized, brand-owned, or attacker — how does fraud prevention tell them apart? The entire fraud stack was calibrated for humans. Behavioral biometrics, typing cadence, mouse-movement entropy, session fingerprinting. None of it survives a language model that fills fields at machine speed, never pauses to read the return policy, and looks exactly like every bot your fraud team was trained to block.

This is the infrastructure gap sitting underneath every agentic commerce pitch. Amazon's entry into the Universal Commerce Protocol and Shopify's AI product discovery both assume a clean handoff between agent and merchant — but neither solves the three-way authentication problem: user-authorized agent, legitimate brand bot, actual attacker. The platform that builds agent credentialing for AI-initiated purchases will own the plumbing of agentic retail. Right now, nobody has it, and nobody at WRC last week was talking about it either.

"Online Shopping Could Be AI's Next Victim" is a deliberately provocative frame and BoF earns the click. The argument: AI-mediated discovery strips the emotional purchase architecture brands spent thirty years building. An agent optimizing for the best V-neck at the right fit score treats scarcity, aspiration, and creator-driven affect as noise rather than signal. Brand becomes a commodity attribute you filter by, not a reason to buy.

I half-buy it. The brands most exposed are discovery-dependent — labels you find through a feed but wouldn't search for directly. Guerlain's first-ever paid influencer campaign is a hedge against exactly this scenario: if an AI filter won't surface a 198-year-old perfume house on its own, you need creator voices that can pierce the intermediary. The brands with genuine product differentiation — Moncler's coat, DeMellier's craft proposition — probably survive an AI purchase layer. The ones selling vibes alone probably don't.

Connection: The fraud prevention gap and this BoF piece are the same story from opposite ends. One is operations — the checkout has no authentication layer for agents. The other is brand strategy — the discovery layer has no emotional hook for an agent that doesn't feel anything. Both describe an industry whose infrastructure was not built for a purchase to happen without a human in the loop.

Bain projects China e-commerce heading toward RMB 1.5 trillion. The framing is bullish; the subtext is saturation. A market triangulating toward a number is a market approaching its expansion ceiling. 证券时报 (Securities Times) is less diplomatic about it in Chinese: Alibaba's "dual engine" has stalled, JD is poaching, Pinduoduo is crushing on price, Douyin is eating the entertainment-first buyer. We read the 380 billion yuan AI commitment as defensive capex; the Chinese financial press reads it as a bet placed by a company that has run out of organic options.

FashionUnited's WRC wrap gives the three-clause industry condition: AI boost, consumer pressure, desire for connection. All true. None pointing the same direction. AI lowers friction; consumer pressure adds cost and compliance overhead; the desire for connection demands physical presence or expensive live social investment. The Congress produced no clean theory of how those three vectors resolve.

VF's Bracken Darrell pitched AI as the brand turnaround mechanism; Authentic Brands pitched an operating system for IP portfolios. Two different theories, both contingent on the consumer pressure part easing — which nobody in Berlin was willing to forecast a timeline for.

Useful numbers in an otherwise late piece: live commerce is 20%+ of total e-commerce in China; the West is sub-5%. This week Topshop — which no longer has stores — paired an immersive catwalk with TikTok live shopping, with Shark Beauty and Lookfantastic as the commerce layer. A brand that no longer has floors using a live format to manufacture the event it can't have in physical retail — that is the Western live commerce story in miniature. Not a native format. A replacement for footfall you lost.

What McKinsey underweights: the gap is partly regulatory. European live commerce operates inside distance-selling rules, cooling-off periods, and return-rate economics that don't apply in China. That's not a technology problem — it's a policy gap — and it means the 20%-to-5% spread won't close purely through platform investment.

Michael Kors has put an AI-powered retail assistant on its website. One brand, one chatbot, no particular novelty — except that Capri Holdings shipped something while everyone else was still in Berlin asking whether to build or buy. The test is simple: can it handle "I need a bag for a January wedding in Edinburgh" without defaulting to the homepage bestsellers list? If it maps intent, it matters. If it can't, it's a press release with a chat window attached.

Alana Pallister sold her house to fund 300,000 Blare tracksuits. She sold a million. The next bet is Coachella as the cultural activation moment for act two. This is the Wonderskin playbook applied to apparel: use the first viral hit to fund the second hook, and route it through an event TikTok will amplify for free.

Connection: Today Sir John covers Forta — WNBA player Lexie Hull's performance-driven makeup brand — and I.AM.GIA runs the mirror route. One is sport crossing into beauty; the other is fashion crossing into athletic silhouettes. The athleisure-to-activewear-to-performance-beauty pipeline is now bidirectional. Brands sitting in the overlap of that Venn diagram are having a better 2026 than the ones staked at either extreme.

Two Shopify moves this week: integrated AI product discovery (intent-reading rather than keyword matching) and Horizon, a new design system built around AI-generated storefronts. Neither is revolutionary in isolation; together they are a claim on the full stack from how a shopper finds a product to how it's presented. This connects directly to the UCP infrastructure fight at the platform layer: if Shopify controls discovery and display for its 2 million merchants, those merchants won't negotiate with Google's Universal Commerce Protocol. They'll ship whatever Shopify tells them works, because they always have.

Kantar's 2026 fashion report introduces "treatonomics" — small luxuries purchased under financial pressure as emotional regulation. Leonard Lauder identified this as the lipstick index in 2001. The new version adds a retail media layer: small treats are impulse-purchasable via in-feed advertising, which is why retail media spend is growing even as full-price apparel is soft. The treat is the category that justifies the ad budget, and beauty understands this better than apparel does.

The data supports it. Mind Games grew 56% without a hero scent. Wonderskin moved 6 million units at a repeatable price point. Today Parallax Pincer covers the beauty bag charm trend — tactile, collectible, small, giftable. Same phenomenon in a different format. The treat is the category; the feed is the store.

Amazon third-party sellers account for 5.5% of UK retail sales — not Amazon itself, but the ecosystem running on top of it. A single platform's merchant layer has become large enough to parse as a discrete retail format in national statistics. For brands sitting out Amazon because it "doesn't fit the positioning": you're leaving a channel the size of a major grocery sub-sector uncontested. We noted the contradiction in Amazon's premium beauty pitch last week; the 5.5% figure explains why brands tolerate the contradictions rather than walking away.

BoF argues the next retail AI frontier is physical: in-store inventory intelligence, staff augmentation, shelf-edge personalization. Companies already executing it: VF's Nedap RFID layer across The North Face and Vans, Dollar General's 12,000-store audio network. What's still missing is the customer-facing layer. Knowing where the size 10 boot is in the back room is operational. Knowing which size 10 customer is standing at the rack — and what she browsed online yesterday — is the harder and more commercially interesting problem, and nobody is close to solving it at scale.

We've seen this movie: Pinterest builds discovery, routes conversion to partners, then spends three years trying to prove attribution to brands that can see the traffic but can't close the loop. The new Chief Shopping Officer's job is to make that proof legible enough that brands write bigger media cheques.

What Pinterest has that TikTok and Instagram don't: purchase intent that isn't time-pressured. A pinned capsule wardrobe waits six months; a TikTok lives 48 hours. That longevity is a genuinely different ad product — not better for impulse, but better for considered purchases: wardrobing, home, travel, occasion dressing. Watch for Pinterest to position more explicitly against Google Shopping than against TikTok Shop. That's the sharper competitive frame and the one that makes the commerce story coherent.

Onton raised $7.5 million for AI e-commerce tooling. Small by US venture standards; notable for category crowding. There are now more funded companies building AI layers on top of existing e-commerce infrastructure than there are clear acquisition paths for them. The mid-market is being approached from Shopify above and from two dozen startups below simultaneously. uMode priced its AI in reais for the Brazilian market; Onton is fishing in a different pond. The question for 2027 is consolidation math, not growth.