Naval Ravikant, co-founder and chairman of AngelList, has a blunt message for investors still betting on pure software companies: stop.

"Pure software is rapidly becoming uninvestable," Ravikant said in a recent episode of the Naval podcast. "If your whole advantage is like, hey, I'm building cool software that other people don't know how to build, I think that's uninvestable."

The reason, he argues, is a two-front assault. Anyone with a clear vision can now assemble a working application in a single session with AI coding agents. And the agents themselves are improving fast enough that within a year, they will likely be producing scalable, well-architected code, not just functional prototypes. 

"The genie is out of the bottle. So if you're a venture investor now, you're looking for hardware, you're looking for network effects, you're looking for AI models," he said.

Must read: AI layoffs may hurt companies too, not just workers: Study warns of ‘automation trap’

Ravikant traces the shift to a specific moment: in late 2025, when Claude Opus 4.5, Anthropic's coding-focused model, landed and demonstrated what he describes as an agent that could "stay on track, build apps soup to nuts, solve thorny problems, and really feels like having a junior programmer at your disposal who's fast, essentially free, and ready to please."

The iPhone in the crosshairs

One of Ravikant's more provocative claims cuts directly at Apple's most profitable franchise. In a tweet from March 23 that he discussed at length on the podcast, he wrote: "AI coding agents can now deliver one-shot, custom apps straight to your phone. It's the beginning of the end for the iPhone's dominance."

The provocation is grounded in a working demo. Ravikant has built what he calls a personal app store, a web interface that lives on his iPhone, where AI agents deliver fully customised applications on demand, piped directly to his device. The apps are keyed to his hardware and can be shared with family and friends, sidestepping the App Store review process.

"If I want a new app that tracks my workouts, I can say: use the functionality of Tonal and Ladder, follow Apple's Human Interface Guidelines to make it look like an Apple app, track my workouts the following way, build me pretty graphs and charts, calculate strength scores, read scientific papers to figure out what the right way to do strength scores by body part is, do a human body diagram, connect to Apple Health for heart rate," he said.

Must read: “It took 9 seconds”: AI agent running on Anthropic’s Claude Opus 4.6 wipes critical database

"I put a lot of it in one prompt and immediately got a working app delivered to my personal app store."

Apple currently prohibits sideloading on iPhones outside the EU. But Ravikant's point is directional: the lock that App Store distribution once represented, the monopoly on reach, is beginning to loosen as personalised, one-shot software generation becomes common. 

Vibe coding: better than video games

Ravikant said that he has been spending two or more hours each night "vibe coding" giving natural-language instructions to AI agents that then write, debug and deploy full applications. He describes it as more compelling than video games, more constructive than doom-scrolling and more aligned with real creative ambition than almost anything else available to a technically literate person.

The analogy to gaming is not incidental. "The way a video game is designed is that it keeps you hooked by giving you feedback and rewards for doing work and it's always at the edge of your capability," he said.

Must read: Don't obsess over coding: Sridhar Vembu's advice to engineers in AI-era 

"But those rewards are fake and the video game is bound. With vibe coding, it's unbounded; you've got a Turing machine running, you can build anything. The objective is created by you and can keep expanding. And it has real-world relevance," Ravikant said.