Ohneis - The Pattern
Tech, AI, design, and the hidden logic behind it all.
Ohneis - The Pattern
The End of Apps: Apple's Invisible Butler Revolution | with Matt
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Technology moves fast, design makes it matter. AI changes everything. This is oh nice. Think back to two thousand seven. Steve Jobs walks onto a stage, pulls a glass rectangle out of his pocket and says There's an app for that. And the world just changed. Fast forward to right now. The average person has eighty apps on their phone. They open maybe nine of them regularly. And every single day they're doing this low grade mental labor, tapping, switching, logging in, tapping again, just to do something as basic as booking a dinner table. Here's the thing. While the rest of us are still doing that, a small slice of tech insiders are already living in a completely different world. They're using AI that doesn't wait to be opened, it just acts. Books the table, orders the car, reschedules the meeting. All in the background while they sleep. Apple knows this is coming, and it's so significant that Tim Cook, one of the most successful CEOs in human history, is stepping down this year to let someone else figure it out. Matt is here with me today, and we're going to tear this apart. The leadership change, the chip war, the death of the app, all of it. Matt, let's start right at the beginning. When you heard Tim Cook was stepping down, what was your first reaction? Hmm.
SPEAKER_01My first reaction was, wait, is this real? Because Cook has been this immovable fixture, right? Like he survived Steve Jobs' death, the AirPods era, the services pivot. But then I started thinking about the timing, and it actually makes complete sense. Because the one thing that's going to define the next decade of Apple, the AI transition, is fundamentally a hardware problem. And Tim Cook is an operations genius, a supply chain wizard. He is not a chip guy. John Turnus, the guy who's taking over, he built the M Series chips, he built Apple Silicon. If there was ever a moment where you need a hardware nerd running Apple, it's right now.
SPEAKER_00And that's the part I keep coming back to. Because from the outside, Apple's 2024 AI rollout, Apple Intelligence, was kind of embarrassing, right? I mean, they announced features that didn't ship. The Siri upgrades got delayed. Tech journalists were basically laughing. So you have this gap. On one side, the general public is still managing 80 app icons like it's 2012. On the other side, developers are already using tools like Clawed Code that can literally browse the web, fill out forms, execute tasks, all automatically. But those tools, they are terrifying to use if you don't know what you're doing. They can click the wrong button, delete the wrong file. So the real question, the question Turnus has to answer, is how do you take something that powerful and make it safe enough for your grandmother? And actually, there's a whole conversation worth having about where current AI still breaks down completely. We did a full episode on that. It's called Why AI is a genius that fails common sense with hope. Genuinely one of the more interesting ones we've done, worth going back to after this. But Matt, back to Apple. Is this a Turnus problem or is it a chip problem?
SPEAKER_01It's both. And they're the same problem. Okay, so here's the thing people miss. When you use ChatGPT on your phone right now, your question leaves your device, flies across the internet to a server farm somewhere, gets processed, and comes back. That whole journey takes time, costs energy, and, most importantly, your data left your hands. Apple's entire brand promise for 30 years has been privacy. What happens on your iPhone stays on your iPhone. The moment they build an AI agent that has to phone home to a cloud server every time it does something for you, that promise collapses. So the Neural Engine, the specialized AI processor they've been quietly building into every chip since 2017, that's not just a performance feature. That is the privacy architecture. That's how an agent can act on your behalf without your data ever leaving your pocket. Turnus understands this at a molecular level. Cook, for all his brilliance, I don't think he felt it in his bones the same way.
SPEAKER_00Right, and there's an analogy I keep using for this. Standard apps are like a tool belt in your garage. They're useful, but you still have to go pick up every tool yourself. An AI agent is like hiring a butler. You say, hang that picture, and the butler goes to the garage, grabs a hammer, does the work, sweeps up. The neural engine is basically how Apple builds a butler that lives inside the phone and never has to leave the house to think. But here's where it gets interesting. Because that same strategy is setting up what I think is going to be the most important corporate rivalry of the next five years. You've got Apple building powerful AI that lives entirely on device. And then you've got Nvidia's Jensen Huang, whose entire empire is built on the idea that AI needs massive cloud servers, massive GPU farms, massive energy. Those two visions are fundamentally incompatible.
SPEAKER_01Yes, and this is the part that doesn't get talked about enough. Nvidia right now is the most valuable company in the world some weeks. Jensen Wong is basically the kingmaker of the AI era. Every tech company is lining up to buy his chips. But if Apple proves that a phone-sized neural engine can run a genuinely useful AI agent without the cloud, that chips away at the entire premise of NVIDIA's dominance. I mean, think about it. If billions of iPhones can handle their own AI locally, that's billions of devices that don't need to rent time on NVIDIA-powered server farms. It won't kill NVIDIA. They do so much more than consumer AI. But it changes the story. And Turnus, as the chip architect, he knows exactly what he's doing. This isn't accidental.
SPEAKER_00I want to push back on something though, because I hear this Apple will figure it out narrative a lot. And sometimes it feels like cope. Like, people say Apple is like a Michelin star restaurant. They wait for others to invent the messy ingredients and then they played it beautifully. And that is true. The App Store wasn't the first app marketplace. Touch screens existed before the iPhone. iTunes wasn't the first MP3 software. Apple almost never invents the raw category. They refine it. But here's my worry. AI agents are not just a new interface layer. They require deep partnerships with the companies that build the actual AI models. OpenAI, Anthropic, Google. And right now, Apple is dependent on OpenAI for series intelligence. That's not a Michelin Star Kitchen. That's ordering ingredients from your competitor and hoping they don't spit in them.
SPEAKER_01That's okay, yeah, that's a fair way to put it. And I don't think Apple would disagree internally. The OpenAI deal was a stopgap. It was Tim Cook saying, We need something on stage at WWDC and we don't have it built yet. But look at the pattern. Apple Maps was terrible at launch, genuinely embarrassing. Now it's excellent and they've reduced their dependency on Google significantly. The same thing will happen with AI models. They are building their own foundation models, slowly, quietly, the way they always do. The question isn't whether they'll get there, it's whether they'll get there fast enough. Because the window where being good enough at AI is a competitive advantage, that window is closing.
SPEAKER_00So let's talk about what this actually looks like for a normal person holding a phone in 2027-2028. Because I think a lot of people hear AI agents and they picture either HAL 9000 or some sci-fi nonsense. What does the practical day change?
SPEAKER_01Okay, so this is the part I actually find most interesting. Imagine you land at an airport. Right now you open Google Maps, then you open Uber, then you check your calendar, then you text your colleague. That's four apps, maybe eight taps, and you're doing all of it while jet lagged carrying a bag. In the agent world, your phone already knows your flight landed 15 minutes late. It's already told the Uber driver. It's already sent your colleague a message saying you're on your way. You pull your phone out and there's just a notification. Your driver is three minutes away. Your colleague knows. Gate B12. You didn't open a single app. You didn't do a single thing. That is a fundamentally different relationship with your device. And I think when people actually experience that for the first time, the home screen of app icons is going to feel like using a fax machine.
SPEAKER_00And that's exactly why the App Store model, which generates roughly $85 billion a year for Apple, is under existential pressure. Because if you're not opening apps, you're not buying things inside apps. You're not seeing app ads. The whole economic engine changes. Turnus isn't just inheriting a cool hardware project. He's inheriting a business model disruption that Apple themselves are triggering. That is a genuinely strange position to be in.
SPEAKER_01Right, and this is something I don't think gets enough attention. Apple has regulatory problems with the App Store in Europe already, right? The Digital Markets Act is forcing them to open up. But the agent's future almost makes that fight irrelevant, because if the agent is the new interface layer, who owns the agent? Who takes the cut when the agent books your dinner? That is the next monopoly fight, and Apple is smart enough to know that if they own the on-device agent, they own the new toll booth. It's just a different toll booth than the App Store.
SPEAKER_00A different toll booth. I like that framing. Because that's what this whole transition really is, isn't it? It's not Apple becoming a different company. It's Apple finding the next choke point. Nat, final thought. If you had to bet, does Turnus pull this off in the next three years? Or does someone else, Samsung, Google, some startup we haven't heard of, get to the everyday AI agent first?
SPEAKER_01Hmm. Three years is tight. I think the honest answer is no single company gets there cleanly. Google has the AI models, but they have a privacy brand problem that makes on-device trust hard. Samsung has the hardware distribution, but they don't control the software stack the way Apple does. Apple has the one thing that actually matters for this to work at a mass scale. They have user trust. When your agent is managing your calendar, your bank account, your messages, trust is the product. And that's Apple's deepest moat. So yeah, if I had to bet, Turnus gets there. Not perfectly, not on time, probably with a few embarrassing stumbles along the way. But he gets there because he's building the vault. And everyone else is still shouting through a megaphone.
SPEAKER_00Matt, that's a great place to end it. The vault versus the megaphone. Thank you for coming on and actually arguing with me a bit. That's what makes this worth doing. If you're listening and this sparked something, subscribe wherever you get your podcasts, leave a review if you want more conversations like this, and share it with someone who still thinks apps are the future. The home screen you're staring at right now, it might be one of the last generations of humans who ever had to use one. See you next time.