Amanda's Take
WWDC Recap: A High-Gloss Keynote, a Low-Key AI Strategy
No big AI leap or surprise hardware. Just a subtle shift back to what Apple does best: design, privacy, and shaping the ecosystem on its own terms.
Programmer Marco Arment recently noted, “Apple no longer wants its executives to be interviewed in a human, unscripted, unedited context that may contain hard questions.” That comment captures the spirit of this year’s WWDC keynote: highly produced with controlled enthusiasm designed to steer the narrative.
Instead of spotlighting major AI breakthroughs, Apple leaned hard into design with its new “Liquid Glass” aesthetic, positioning it as a major innovation that blends hardware and software and promotes continuity across platforms. It felt a little over-hyped at times. But after last year’s lofty promises about AI that failed to materialize, Apple seems to be retreating to what it knows best: thoughtful, user-friendly design. As Stratechery’s Ben Thompson put it, this was a “retreat to safety.”
But while Apple Intelligence was glossed over during the keynote, aside from a flashy image playground (do our emojis really need sequin jackets?) and live translation features, the notable AI updates came later. In the Platform State of the Union, Apple quietly announced its new Foundation Models framework: free, on-device LLMs that let developers experiment with privacy-friendly AI. Apple also expanded its integration with OpenAI and added support for custom models in Xcode.
But it’s not lighting a fire in the developer community. As one developer told the Wall Street Journal, “I’ve not really felt the intensity of excitement around what Apple is releasing.”
On the privacy front, Apple also quietly upgraded AdAttributionKit, adding features like flexible conversion tags and customizable attribution windows. Meanwhile, Safari stepped up its advanced fingerprinting protections, blocking access to device signals and URL parameters by default. Branch just published a more technical breakdown for marketers, but the tl;dr is that modeled attribution is more relevant than ever.
The message is clear: privacy is still king, on-device is the way, and “delightful” design is front and center. For developers and marketers, that means playing by Apple’s rules, designing natively, and embracing privacy-first, on-device tech.
Intro
Trends in AI
Throughout the last decade, the name Mary Meeker harkens memories of an incredibly exciting time in mobile. Every year, she produced hundreds of up-and-to-the-right slides showing incredibly fast, promising, and lucrative growth. And the mobile industry would collectively smile, shake our heads, and pat ourselves on the back for being part of such an incredible story.
After going dark in 2019, she’s returned — leading her own firm, Bond Capital — with 340 pages of AI trends that make mobile’s growth look prehistoric. The TL;DR? Unprecedented (a word she uses 64 times).
The scope is too vast to summarize here, but one taste: OpenAI predicts they will run autonomous companies and simulate human minds — all within a decade.
Too much data? Cleanse your palate with Sam & Jony’s awkwardly charming $6B “Would You Be My Friend?” video, or spiral into existential dread with Yuval Noah Harari, warning about our C‑minus response to the Industrial Revolution.
Interesting Reads
Luma's State of Digital 2025
Luma Partners just released their annual trend report, and — as always — they distill the complex media market into clear, actionable insights. This year, they highlight three key themes:
- “Your services are my opportunity:” A new hierarchy of AI applications is emerging, poised to deliver incredible value, and setting the stage for a battle between AI-native challengers (e.g., OpenAI) and incumbent tech giants (e.g., Google).
- Outcomes > Proxies: AI is enabling performance media to optimize for real business outcomes — not just vanity metrics.
- Media consumption is evolving: UGC creators drive most of the $267B in video ad spend, but capture just 10% of the revenue. This imbalance signals a major opportunity for technology to unlock value.
Industry Buzz
Nike’s Return to Amazon May Yet Prove a Mistake
Following slumping sales and an almost 70% decrease in stock value from a 2022 peak, Nike is abandoning its four-year-old, direct-to-consumer approach and (re)embracing the wholesale model.
It says a lot about the difficulty of performance advertising when the world’s most recognizable logo stumbles so blatantly putting it into practice. Former Nike executive Massimo Giunco gives an interesting analysis of their failure and while his view on performance media is reductive and flawed, his point about Nike turning its back on a thriving wholesale marketplace and opening up a path for competitors to steal marketshare is hard to argue against.
Looking ahead, the relinquishment of a direct-to-consumer approach may cause a ripple effect that hurts Nike during the next emerging paradigm shift in advertising: AI.
As brands struggle to ready themselves for AI, increasing wholesale channel reliance surrenders key important components for success in AI: owning customer data and having the ability to quickly measure and iterate in the new medium. Allowing Amazon — and other retailers — to own the customer experience and purchase process won’t do Nike favors in this new medium.
Spotlight
Google IO v. WWDC
In stark contrast to Apple’s WWDC, Google’s yearly IO was awash with jaw-dropping displays of technology (most of which will, unfortunately, never receive scaled deployment).
On-device LLM v. Deep Search: Apple’s on-device LLM (6 min) is being opened up for third-party apps (wait, Apple has an LLM?) meanwhile Google’s updates to Gemini, include impressive search, deep research, and personalization enhancements.
Visual Intelligence vs Virtual Try-On: A key feature of Apple’s Visual Intelligence (38:50) allows you to search the web for items in a screenshot. (If this sounds familiar, it’s because it used to be called "Google Image Search" launched in 2001. Meanwhile, Google allows you to virtually wear an item.
Genmoji v. Veo: Apple’s Genmoji allows you to now combine two emojis (!). Meanwhile, Google Veo allows you to generate shockingly good video with a prompt. Yes, it’s really that amazing.
Vision Pro OS vs Android XR: Apple announced (1:06) persistent widgets for Vision Pro (think a virtual wall clock), meanwhile Google announced real-world AR glasses allowing virtual overlays and real-time interaction with Gemini (think: what coffee is that person drinking or how do I get there?)
Admittedly, despite the overwhelming technological awe of Google’s very exciting technology, in practice, Google does have a habit of abandoning products, so the likelihood of these releases changing the world is small. Whereas, Apple’s technology is generally ready for prime time and will be on your device within the month.