Codes of Culture Issue 96
EssilorLuxottica's bet on eyes as a health platform
Welcome back to Codes of Culture.
This week’s issue is slightly longer with 10 stories on our radar covering recap of the recent Shoptalk event in Las Vegas where the best technology solutions present to the biggest retailers. Agentic commerce has been at the top of everyone’s agenda, and tracking announcements at events gives us a macro understanding of the trends and capital flowing into the space.
We also recap both Prada’s Frames symposium, which highlighted the existential question of our times: who owns AI image generation, and EssilorLuxottica’s wearables summit, which interestingly positioned the eye and its wearables as a health platform that tracks data and uses AI to detect systemic conditions.
Especially interesting (and worrying) this week was The New Yorker’s coverage of Sam Altman’s growing influence on the future of AI and whether one person can really be trusted with all the control. Scroll on for all the stories.
Finally, we have officially re-launched the newly updated Future+ website, which has been a labour of love since the start of the year.
If you are new here, or want to catch up on the best of Codes of Culture, we recommend you start here and remember to subscribe for full access to our news, insights, podcast and global events.
📖In this issue:
Google’s AI-powered adverts drove an 80% increase in sales for Aritzia.
Prada Frames explores who controls imagery in the age of AI at Milan Design Week.
Pixel Moda demonstrates how AI is revolutionising large-scale fashion imagery production.
Nothing announces its entry into the AI glasses market with a design-led approach.
EssilorLuxottica highlights the convergence of eyewear, AI, and healthcare innovation.
Medvi faces scrutiny after revelations of fabricated data behind its AI success story.
NASA’s Artemis II mission advances human spaceflight further, pushing lunar ambitions.
Agentic commerce owns discovery. The storefront still owns the close.
The defining story at Shoptalk Spring was OpenAI quietly sunsetting Instant Checkout, its attempt to complete purchases inside ChatGPT, and pivoting to a discovery-and-handoff model through its Agentic Commerce Protocol, with Target, Sephora, Nordstrom, and Wayfair already integrated. Sephora is now inside ChatGPT. Gap is purchase-enabled inside Gemini. Meta has rebuilt shopping summaries and checkout across Instagram and Facebook. The floor data was sobering: 60% of shoppers who receive an AI recommendation still go and research before buying; only 18% transact immediately. That gap is the whole story. AI is taking over the moment of intent, but it is not yet closing the sale. For retail and brand operators, the question is whether your product feeds, pricing logic, and data infrastructure are clean enough to convert attention that now arrives from an interface you do not own.
The New Yorker is asking the governance question OpenAI keeps deferring.
Drawing on new interviews and internal documents, The New Yorker’s profile revisits the 2023 OpenAI board crisis and lands in an uncomfortable place: the doubts about Altman were never really about personality. They were about candour, the concentration of power, and whether the people constructing this technology are sufficiently accountable for what they produce. OpenAI has spent two years managing that story. The New Yorker is reopening it. For capital allocators and enterprise buyers with material exposure to OpenAI’s ecosystem, the governance question is not an abstract one. A single founder’s control over a general-purpose technology at this pace and scale is one of the most significant structural risks in the market right now. The profile is asking it on the record. The answer is still open.
Google’s AI ad tools drove Aritzia’s revenue up 80%. Performance marketing has a new architecture.
Google’s AI Max campaign tool reads a brand’s creative assets and product catalogue, then interprets conversational search queries, now averaging 45 words against the old two-to-three keyword standard, to match intent to inventory in real time. Aritzia’s 80% lift in online revenue is not a creative optimisation story. It is evidence that the matching layer between customer intent and product is being rebuilt by the platforms that own the search interface, and is being rebuilt fast. For fashion and retail brands with significant paid acquisition budgets, the question is not whether to engage with these tools. It is whether your creative assets, product data, and catalogue structure are clean enough to perform inside them. The brands that are waiting are not just missing out on efficiency gains. They are being outpaced on discoverability by those that are not.
Prada Frames is asking who controls the image.
The fifth edition of Prada’s annual symposium, In Sight, held at the Santa Maria delle Grazie complex in Milan, was focused on image-making as a political and material force: the infrastructure it depends on, the energy and labour it consumes, and who controls the conditions of its production. Conceived by Formafantasma, the event spent five editions building toward exactly the question AI image generation has now made unavoidable. What is striking is where the conversation has arrived. This is no longer a cultural theory seminar. The struggle over who owns visual culture and who controls the systems that produce it is showing up in boardrooms, in content contracts, and in procurement decisions. Prada is staging that conversation in one of the most symbolically loaded spaces in Milan. That is not incidental.
Pixel Moda produced 14 million fashion images with AI.
Pixel Moda creates AI-assisted and AI-generated content for more than 900 global brands, and presented its model at BoF VOICES alongside Etro’s chief executive, who credited the partnership with 46% e-commerce growth over 12 months. The model is deliberately human-led: AI directs photographers, stylists, and models on set, then uses those assets as inputs for generative outputs, including body swaps, background changes, and per-SKU video, content that used to be cost-prohibitive for all but the largest houses. The signal here is not that AI is replacing the studio. It is that the studio now has a different architecture, and the question is whether the creative layer directing it is strong enough to maintain brand identity as volume multiplies. The brands still running workflows designed for 500 images a season are already operating in a different era than the ones that are not.
Nothing is entering AI glasses.
London-based Nothing is developing AI glasses targeted for the first half of 2027, featuring cameras, microphones, and speakers in a design similar to Meta’s Ray-Ban glasses. CEO Carl Pei shifted from initial resistance to a multi-device view after seeing the category’s early traction, itself a signal worth noting. For luxury brands and hardware operators tracking ambient AI devices, Nothing’s entry matters for the same reason Samsung’s partnership with Gentle Monster mattered: it confirms that the primary constraint on mass adoption is design and social acceptability, not the technology. Meta, Samsung, Google, and Nothing have now converged on the same hardware thesis within a narrow window. When operators of that scale reach the same conclusion at the same time, the question for luxury is not whether eyewear strategy belongs in the conversation. It is how quickly it moves to the centre of it.
EssilorLuxottica is repositioning the eye as a health platform.
EssilorLuxottica’s inaugural SWITCH summit convened more than 1,000 doctors, scientists, and technologists around oculomics: using advanced retinal imaging and AI to detect systemic conditions, including diabetes and neurological disease, through the eye. The company already controls the dominant global eyewear distribution infrastructure through Ray-Ban Meta, Oakley, LensCrafters, and Sunglass Hut. That combination, diagnostic capability sitting inside a distribution system that already reaches hundreds of millions of people, is the thing worth paying attention to. For luxury wellness, longevity, and premium healthcare investors, the category is repositioning from optical correction to continuous health monitoring. The interface closest to the face is becoming the one that reads it most intimately. What sits above that layer, in experience design, data governance, and brand relationship, is still wide open.
A startup in Surat is pricing McKinsey-style strategy reports at $250 a month.
Rocket, based in Surat with operations in Palo Alto, generates consulting-style strategy documents covering pricing, unit economics, go-to-market, and competitive intelligence, drawing on more than 1,000 data sources. Its thesis is direct: as coding is commoditised by AI, the strategic layer above it is where the next value gap opens. The outputs still require human verification. But the more consequential point is the price signal itself. The market is beginning to set a floor for AI-assisted strategy work, and that floor is lower than most incumbents have modelled or are comfortable acknowledging. The firms most exposed are those whose value proposition rested on access to proprietary frameworks and aggregated data rather than judgment built from genuine operating experience. That distinction between analytical access and earned perspective is where the value will concentrate and where the pressure will fall hardest.
Medvi was celebrated as the first one-person unicorn, but it was built on 800 fake doctors.
The New York Times profiled Matthew Gallagher as proof that one person with AI could build a billion-dollar company. The revenue figures cited, $401 million in 2025 from two employees, came from an unaudited private LLC with no independent verification. What has been independently documented is rather different: more than 800 fake Facebook doctor profiles with AI-generated images, deepfaked before-and-after patient visuals, and an FDA warning letter for misbranding violations that arrived six weeks before the profile ran. A class action and a data breach affecting 1.6 million patient records followed. The Times piece has since been heavily scrutinised. For investors and enterprise buyers tracking AI in healthcare and wellness, Medvi is the clearest illustration yet of what happens when AI is used to fabricate the trust layer rather than earn it. The regulatory window that made this possible is closing. The harder question is how many companies in adjacent categories are operating closer to this model than their positioning suggests.
Artemis II has taken humans farther from Earth than any crew in history. The infrastructure for a lunar economy is being tested now.
NASA’s Artemis II crew has completed a circumlunar flyby, surpassing the distance record set by Apollo 13. This is not a landing, but that is precisely the point. The strategic significance lies in what the mission is building toward: Artemis III targets the lunar south pole, with the longer ambition of sustained human presence beyond Earth. For deep-tech investors and capital allocators tracking the space economy, the Artemis programme is public infrastructure in the most literal sense: the enabling layer on which commercial lunar operations, resource extraction, and eventually off-Earth compute will depend. The test is live. The commercial layer above it is being built in parallel. The organisations that understand the infrastructure timeline are the ones positioned to shape what eventually runs on top of it.