Spyglass Dispatch: Apple's iPhone "Problem" • End of GPTs? • AR in Your Ears • Gemini in Your Pocket • Amazon's Grocery Confusion • How to Write About the Election

The Spyglass Dispatch is a free newsletter sent out daily on weekdays. Feel free to forward it on to others who can sign up and view previous dispatches here. You can also view this dispatch on the web here.

Saying something will never happen is dangerous. Never is a long time. In fact, never is forever. And yet, it feels impossible right now to think that Apple will ever create a new product as big as the iPhone. And they're seemingly admitting that possibility as well in their updated 10-K filing. Sure, Services may one day pass the business in terms of revenue, but again, it's hard to see a single product doing so. I mean, maybe in a couple decades if you only count revenue and don't take inflation into account? Hell, that's what Hollywood does for their "records", so why not? But even then, I'm not sure something can top the iPhone! That's how good of a business the iPhone is and remains. The problem I wrote more about today...

Apple Admits the Obvious

There’s no topping the iPhone. As a business, there may never be.

Spyglass •M.G. Siegler

I Think...

🤖 OpenAI Shifts Strategy as Rate of ‘GPT’ AI Improvements Slows – There have been growing whispers for months that we were hitting a plateau when it comes to LLMs. But this doesn't mean the era of AI advances is over, it just means it's morphing. For a while it seemed like synthetic data would solve the data problem and now newer "reasoning" models seem to be alleviating some of ouroboros concerns there. More audio and video inputs through "agentic" tools should help as well, but at some point, "real world" data from devices out and about will probably be needed too. It all points to a world in which one model doesn't rule them all, but many models are used in concert depending on the job to be done. So it would be a fitting way to evolve from the "GPT" era if OpenAI itself were to morph "Chat" formally into just one tool in their toolbelt. More generally, I continue to think this "slowing down" will help AI from a product-perspective with the ground on which to build finally stabilizing a bit. [Information 🔒]

🗺️ With AI, the Future of Augmented Reality is in Your Ears – By his own admission, Dennis Crowley keeps building the same thing: first with Dodgeball, then with Foursquare, and now with Beebot (the first product from his Hopscotch Labs – clever). And actually Beebot itself is just a version of a product called 'Marsbot' which was a project inside Foursquare that was DOA because of horrible timing with the pandemic. But as a firm believer in just sticking around long enough to give yourself a shot to be "right place/right time" perhaps now is that right time with the introduction of AI mixed with the ubiquity of AirPods (remember Field Trip and Highlight?). I also appreciate his notion that while the future may be a true multimodal AR, you can't just sit around and wait for the devices that will deliver that, as you might be waiting for a decade. Start now with audio and make it extensible enough to morph into visuals, etc. For now, "Marauder's Map for AirPods" sounds fun. [Crazy Stupid Tech]

🌟 Standalone ‘Google Gemini’ App Spotted for iPhone – While I quite like the Google iOS app in which it's baked right now, Gemini is probably at least one click too far away if they want to compete with ChatGPT, Claude, Perplexity, etc – all of which have their own purpose-built apps. Microsoft's Copilot is also now decoupled from the Bing app even though it's still the central part of that app. Also, given the way we can now see how the ChatGPT integration works in the new iOS betas, it looks like the way to set-up other AI services to pipe into Siri will be through their apps. Also: Gemini Live, their voice AI, is baked into this version, which wasn't true in the Google app. Also, also: Google Assistant still has its own app which I'm sure won't confuse anyone. [9to5Google]

🛒 Amazon Targets Delivery Operations in Bid for Bigger Grocery Business – Clean up, Aisle Amazon. I mean, what a mess. In a bid to catch up with Walmart in grocery, they're going to start fulfilling some Whole Foods online orders through Amazon Fresh warehouses, which is the other grocery brand they started after spending $14B on Whole Foods. But there's also now 'Amazon Grocery' which is another new store and brand. And none of those are 'Amazon Go' What chat apps are to Google, grocery stores are the Amazon. You want more people to shop (and trust) your grocery store online? Step one: pick one brand. Or even two would be a huge improvement at this point! [WSJ 🔒]

💍 How It Went – A poignant post from John Gruber. I honestly postponed reading it for a bit as I was sick of reading anything about the election. But the election is a minor player here – wrapped in this beautiful bow of a story which doubles as a dose of perspective. Well worth your time. [Daring Fireball]

I Link...

• For all the talk of AI disruption, it does feel like Chegg is really feeling it. The stock is off 99% from its all-time high – a lot of that is just a pandemic comedown, but that would-have-been OpenAI partnership on "Cheggmate" is interesting. What happened there?! [WSJ 🔒]
• As Firefox turns 20, Frederic Lardinois sat down with Mozilla's (interim) CEO Laura Chambers to talk about where they're focused going forward: privacy, AI, and seemingly Europe, given the favorable DMA elements. She doesn't sound overly concerned about the DoJ ending Google's default search payments, but she's undoubtedly downplaying it. Pulling for Firefox, but not a great position to be in, obviously. [TechCrunch]
• Interesting profile of John Janick, the CEO of Interscope Records – who clearly runs it very differently than Jimmy Iovine did, but no less effectively in terms of securing talent, it seems. [FT 🔒]
• Intel's gaming performance issues with their 'Arrow Lake' chips were "self-inflicted." Not great news at a not great time for Intel. Luckily for them, it's not like the new 'Copilot+ PCs' are doing any better in this regard. [HotHardware]
• True end of an era with Henry Blodget fully leaving Business Insider, the publication he started 17 years ago. He's says he's going to write fiction – though I can't quite tell if that's just an Elon joke. [TalkingBizNews]
• Since not every story has J.R.R. Tolkien creating entire languages for a narrative, "conlangers" (language constructors) come in to help build out worlds for properties like Dune and Avatar and Game of Thrones. [NYT]
• Mark Gurman quasi-confirms an M5-powered Vision Pro is in the works "assuming it's released". As previously noted, while Apple messed up the strategy overall here, this is clearly the path forward until they can get a model that's priced low enough to actually sell in volume. Also, an M5 version would allow them to do a lot of AI work for the device. And the new "ultrawide display" in the visionOS 2.2 beta sounds like an actual winner, software-wise. [Bloomberg 🔒]
• Amazon may also be working on their own newfangled AR/AI Smart Glasses – but not the Alexa/Echo variety, these are to help their delivery drivers in the last mile of package journeys. [Reuters]
• Well, this sure feels like they're at least marketing it as the last Mission: Impossible movie. Though Tom Cruise refuses to commit to that. The trailer is what you'd expect: a lot of running, a lot of hanging from planes. [THR]

I Quote...

"They mostly end up with expensive urine."

-- Dr. Patrick Burns, a clinical associate professor at Stanford, talking about the rise of startups selling various concoctions focused on electrolytes.

Mainly, I'm upset that this entire 1,500-word post lacks a single reference to Brawndo. I mean, Brawndo's got what plants crave, it's got electrolytes.