ANDAM — the French fashion establishment's prize committee — awarded its 2026 Prix de l'Innovation to Alphalyr, a platform that does demand analytics and sell-through forecasting. Not a generative AI catwalk tool. Not an avatar dressing room. A data platform that tells buyers what to reorder and when. The trade press will frame this as France embracing AI. I'd frame it as ANDAM catching up to what Japanese convenience chains worked out a decade ago. Sir John writes today about Japan's retailers putting AI where the margin actually is — reordering. The French fashion establishment, in its elegant way, just agreed that this is the correct problem to solve.

The Securities Times (证券时报) doesn't hedge: Taobao's "dual engine" growth has stalled, JD is poaching Alibaba engineers, Pinduoduo has turned pricing into attrition warfare, and Douyin has eaten the discovery layer from underneath. The ¥380 billion AI infrastructure pledge reads in this framing as a holding action by a company that knows it's losing ground and is buying time. We covered this from Alibaba's side last week — the defensive capex thesis. The Chinese financial press makes the same argument in cruder language: the rivals are not waiting for Alibaba to find its footing. Meanwhile Sir John writes today about Pinduoduo policing AI content while the West hasn't started — which shows how far ahead of Western platforms Pinduoduo's operational thinking is already running.

World Collective launched what it's calling the first Digital Product Passport-ready textile library: a structured database of textile properties brands can link to their DPP records as the EU mandate approaches. This is precisely the infrastructure gap Neritus Vale diagnoses today in The Digital Product Passport Turns Sourcing Into a Data Problem. The library is live. Brands still have to populate it, which means walking the supply chain with a verification process and writing up every fiber origin, treatment, and certification. That part has not been automated. The bottleneck in DPP compliance was never the standard — it was always catalog accuracy, and a pre-built library is only as good as the data someone enters into it.

Cartier's owner posted €22.4 billion for the full year, citing "geopolitical shifts" without naming the China softness and Gulf resilience behind the number. The result is directly relevant today because Sir John writes about luxury AI being pointed at the biggest spenders, not the masses. Richemont's margin structure is the argument in favor of that choice: when the top 2% of clients generate a disproportionate share of annual revenue, every AI dollar allocated to the other 98% is a provable misallocation. The interesting question is whether their client profiling can identify the incoming top 2% before competitors do — Cartier has the heritage, the question is whether it has the data infrastructure to match.

Same week, Sir John writes about licensed goods outgrowing retail because the IP is the product and Gulf capital buying Cavalli as pure IP. The three together make one argument: value in fashion has migrated away from the manufactured object and into the brand asset itself. Richemont has known this for thirty years. The rest of the industry is just arriving.

German trade publication etailment has the most useful agentic commerce piece I've read this month and almost nobody outside Germany will read it. The argument: the entire fraud prevention stack was calibrated for human behavior. Behavioral biometrics assumes a human moves the mouse. Device fingerprinting assumes a human owns the device. Velocity checks assume a human can only buy so fast. An AI shopping agent — or an attacker mimicking one — breaks all of these signals simultaneously. The question Pascal Zangerle raises is sharp: when a checkout fires, who authorized it? The human who configured the agent last week? The agent's decision model? The platform routing the order?

This is live today for anyone running a shopping agent on Shopify or Amazon. We noted last month that retailers deployed agentic apps faster than evaluation benchmarks were built. The fraud infrastructure is even further behind. When Neritus Vale writes today about Radar's $170M raise and the race to own retail's intelligence layer, agent authentication is the component of that intelligence layer that nobody has claimed yet. Whoever claims it first will have a toll on every agentic transaction in retail.

Glossy's check-in on the two beauty AI platform bets finally gives us side-by-side material. Sephora is inside ChatGPT's shopping layer; Ulta went with Google Gemini. These are not equivalent bets. ChatGPT shopping integrations are discovery-oriented — someone asking what concealer to buy. Gemini is more tightly coupled to Google Search and Google Shopping, which means Ulta is essentially buying an enhanced SEO position dressed in AI clothing. Sephora is betting on the conversation layer. Both are defensible theories of how discovery happens; they just describe different customers. We noted in May that every beauty retailer was converging on the same curation pitch — these platform partnerships are where the actual differentiation is playing out, which makes the platform chosen more revealing than the press release. And since Sir John writes today about how the dupe economy finally reached Guerlain, it's worth noting: the discovery layer these platforms control is partly about whether the first result surfaces the original or the copy.

Hims & Hers will "soon" launch an AI companion for weight-loss users. The press framing is all support, journey, accountability. The business logic is GLP-1 subscription retention and prescription refills. An AI that checks in on medication adherence, nudges you to log meals, and asks how you're feeling is doing CRM work that used to require call center operators — but at near-zero marginal cost. Early digital therapeutics research suggests AI-assisted adherence can improve protocol completion rates at scale. Whether it works at Hims & Hers' cost structure is the test. The AI companion story is also the best framing they have for turning a prescription service into a platform — once you hold the platform, the next product doesn't need a doctor's note.

Alphabet launched its AI Health Coach on May 19. Glossy frames this as Google entering the wellness race. The more precise frame: Google solving its post-cookie targeting problem with something exponentially better than cookies — longitudinal health data. What you eat, how you sleep, whether you exercised last Tuesday — those signals predict your next supplement, skincare, and athletic apparel purchase more accurately than any behavioral retargeting ever could.

The connection to the Hims & Hers story above is worth making explicit: both platforms are chasing the same wellness data layer that eventually determines what you buy across every adjacent category. Google's advantage is search integration; Hims & Hers' advantage is the prescription relationship. Neither has the whole picture yet. Whoever assembles it will hold the intent signal for a significant share of fashion and beauty purchasing — before you know you're going to buy anything. That's more valuable than any third-party cookie ever was, and the regulatory apparatus that killed cookies has not yet caught up to this version.

Pepco Group — Poundland, Pepco, Dealz — posted 5% revenue growth and 52% profit-after-tax growth in H1. That gap is operational leverage when you run discount retail with cost discipline: no AI-driven personalization platform, no loyalty program that costs €20 million annually to maintain, no sustainability pivot requiring a government subsidy. The right five aisles, right price, right store format. Meanwhile Sir John writes today about Primark and H&M lobbying governments to subsidize resale — the value end of fashion is splitting between Pepco's ruthless simplicity and the legacy fast fashion players asking the state to fund their pivot. One of these strategies produced a 52% profit jump. The other produced a press release.

Walmart's Q1 FY27 commentary named Flipkart and Flipkart Minutes — its quick commerce arm — as contributors to both ecommerce and advertising growth. Walmart bought Flipkart in 2018 for $16 billion; for most of that time it looked like a patient strategic bet not yet producing US-comparable numbers. The advertising growth signal is the most interesting disclosure: Flipkart is becoming a retail media network, and Walmart is learning from India how to run one at scale. We noted in May that the Gulf had already run the MENA playbook McKinsey was still recommending — India is doing something similar to Walmart. The operating model built for Flipkart is now teaching the parent what US retail advertising looks like in five years.

Amazon marketplace sellers — not Amazon itself, just the third-party tenants — account for 5.5% of all UK retail sales. That single number makes the "Amazon is the mall" framing obsolete, because no mall accounts for 5.5% of national retail. Amazon does. Its tenants pay rent in fulfillment fees and advertising at rates that would make any commercial landlord envious. The logistics surcharge and network opening we covered earlier this month reads differently at this scale: Amazon isn't just setting fees for sellers, it's setting the operating cost for one in twenty pounds spent at British retail. And the Glossy Beauty Podcast asking this week whether a big beauty sale is Amazon's play for prestige credibility is the wrong question — twelve Estée Lauder brands landing on the marketplace suggests Amazon wins beauty through distribution, not campaigns.

Brigette Pheloung renamed the charter jet "Acquired Air," flew sixteen guests to her bachelorette, and generated launch heat for her sister's Phe Phe fashion collection in the process. This is the third bachelorette-as-brand-launch story this week. Swan Beauty's St. Barth's trip broke the internet; here's fashion doing the lower-budget version. Refy gave up approval rights entirely and sold out at 200% of forecast; Parallax Pincer writes today about how The Devil Wears Prada 2 is a hall of mirrors of co-marketing. The bachelorette trips are the founder-led, overhead-minimal version of the same impulse. The playbook is now legible enough that its saturation clock is running. Eighteen months, maybe less, before the backlash pieces start.

The BoF opinion piece arguing that AI could be online shopping's next victim is worth reading slowly, even though the framing is wrong. The actual argument is that agentic shopping — where an agent completes the purchase for you — eliminates the browsing experience that made e-commerce interesting and emotionally satisfying. Discovery dies. Efficiency wins. The piece is right about the emotional economy of browsing. It's wrong to name AI as the villain. The villain is time poverty. AI agents are the rational response to not having forty minutes to compare hoodies across six sites.

What actually dies when agents take over is the impulse-buy and the rabbit hole — which is genuinely bad for platform players like Pinterest who monetize pre-purchase wandering, and for TikTok Shop, whose comment section is the closing mechanism. It also means the discovery-driven breakout is harder: I.AM.GIA sold a million tracksuits through social browsing — an agent executing against your preference profile would never have found I.AM.GIA in the first place. That's the real cost. Not shopping dying. Novelty dying.

Irisphera — visual discovery and online fashion experience platform — announced an Oracle partnership to position itself as an enterprise AI player. This is the second major fashion-AI/enterprise-cloud pairing this month; OTB put five luxury houses on Google Cloud's AI layer earlier in May. The pattern: fashion-AI startups attach to Oracle, Google, or SAP to close enterprise accounts they couldn't close on technology alone. The Oracle badge is the sales instrument. Irisphera's actual technology — visual discovery, catalog enrichment — is sound. The risk, as with OTB, is vendor dependency dressed as partnership. Watch which of these deals include data portability clauses. The ones that don't are the ones worth scrutinizing.