Highlights
Naval Ravikant, the co-founder and chairman of AngelList, delivers a straightforward message to investors who are focused on pure software companies: it is time to reconsider.
Ravikant conveys that pure software is swiftly becoming a challenging area for investment. He mentioned in a recent episode of the Naval podcast that if an investor’s primary advantage lies solely in developing innovative software that others cannot create, it is increasingly seen as an unwise investment.
The basis for his view stems from a dual challenge. Anyone with a clear vision can now create a functional application in one session using AI coding tools. Additionally, these AI tools are evolving rapidly, and it is likely that within the next year, they will be generating scalable, well-structured code rather than just basic prototypes.
“The genie is out of the bottle. Therefore, if you are a venture investor now, you should be focusing on hardware, network effects, and AI models,” he noted.
A Shift in Perspective
Ravikant identifies a pivotal moment for this transformation: in late 2025, when Claude Opus 4.5, a model focused on coding from Anthropic, emerged and showcased an agent capable of effectively managing app development from start to finish, addressing complex issues, and resembling having a junior programmer who is fast, virtually free, and eager to assist.
The iPhone Under Threat
One of Ravikant’s more striking assertions challenges Apple’s most lucrative product line. In a tweet from March 23, which he elaborated on during the podcast, he stated that AI coding agents can produce custom apps directly for smartphones. He believes this marks the decline of the iPhone’s supremacy in the market.
This assertion is backed by a practical demonstration. Ravikant has created what he describes as a personal app store—a web interface on his iPhone that allows AI agents to provide fully personalised applications directly to his device. These apps are designed specifically for his hardware and can be shared with friends and family, bypassing the App Store’s approval process.
“For instance, if I require a new app to monitor my workouts, I can instruct the AI to utilise the features of Tonal and Ladder, adhere to Apple’s Human Interface Guidelines for aesthetics, track my workouts in a certain manner, create appealing graphs and charts, compute strength scores, consult scientific literature on optimal strength scoring methods by body part, produce a human anatomy diagram, and integrate with Apple Health for heart rate tracking,” he explained.
“By inputting this extensive prompt, I instantly received a functioning app delivered to my personal app store.”
While Apple currently disallows sideloading on iPhones outside the EU, Ravikant’s argument points towards a significant change: the stronghold that App Store distribution once had, monopolising market access, is beginning to weaken as customized, one-off software creation becomes more widespread.
Engaging with Vibe Coding
Ravikant mentions that he has been spending two or more hours each evening ‘vibe coding’, which involves giving natural-language commands to AI agents that then write, debug, and launch complete applications. He finds this experience more engaging than video games, more productive than mindless scrolling, and more fulfilling in terms of genuine creative expression compared to many alternatives available to skilled individuals.
The comparison to gaming is no coincidence. “Video games are constructed to maintain player engagement by providing feedback and rewards for effort, always pushing the limits of capability,” he indicated.
“However, these rewards are artificial and the video game experience is confined. In contrast, vibe coding is limitless; with a Turing machine at your disposal, anything is possible. You determine the objectives, which can continuously evolve, and the creations have tangible relevance in the real world,” stated Ravikant.
