Kevin Rose on social media's death by AI bots, the future of personal software, and why CS degrees may be obsolete

Summary

Kevin Rose, partner at True Ventures (managing $4 billion in assets), sees the convergence of AI coding tools and vibe coding as a structural shift that reorders who can build software and who controls the economics of doing so.

Personal Software Is the New SaaS

Rose argues the most consequential near-term disruption is the rise of personal software. Tools like Cursor already allow non-engineers to function at a senior coding level, and within six to twelve months he expects that capability to reach mainstream consumers. The implication is a massive expansion in the number of apps and products hitting the market, built by founders who never need to raise a seed round.

Founders reaching 100,000 to 500,000 users and meaningful ARR before ever speaking to a VC will become routine, Rose contends, compressing pre-money valuations upward or eliminating fundraising entirely. Lifestyle businesses generating $20 million in ARR without institutional capital become a realistic outcome. That dynamic is structurally negative for early-stage software VCs.

One practical illustration came from within the conversation itself. A media company described building a fully custom show-management platform and a bespoke ad-tracking system, software with a total addressable market of exactly one. The economics only work because AI-assisted development has collapsed build costs, not because the software could ever be commercialized.

Vibe Coding's Real Bottleneck Is Maintenance, Not Prototyping

The current constraint in AI-assisted development is not initial build speed but long-term code maintenance. A ten-hour build can generate hundreds of hours of follow-on debugging work. Rose believes that problem is close to being solved, pointing to Vercel CEO Guillaume Raynaud's concept of "vibe check," an agent that autonomously identifies and fixes bugs at scale. Because software bugs are defined problems with deterministic answers, Rose argues agent-based maintenance is tractable in a way that, say, drug discovery is not.

Hardware Is Where VC Capital Still Matters

As software capital requirements shrink, Rose identifies hardware as the category that still genuinely needs institutional funding. True Ventures backed Ring, Fitbit, and Peloton, all of which required hundreds of millions of dollars to reach scale. That dynamic has not changed.

On AI wearables specifically, Rose is skeptical of always-on devices, calling them "insanely creepy" and arguing that any product that makes bystanders want to react physically is a signal not to invest. He sees more promise in unreleased devices designed to balance extended functionality with privacy. He is more constructive on health-monitoring hardware, citing his board tenure at Oura, where he helped build the biohacker ambassador network including Tim Ferriss, Peter Attia, Ben Greenfield, and Dave Asprey and worked with UC Berkeley sleep researcher Matt Walker to improve the ring's sleep algorithms. He expects AI to unlock additional health signals from such devices, including blood pressure monitoring and eventually integration with individual genomic data.

CS Degrees Are No Longer the Default Recommendation

Rose says he spent roughly twenty years telling young people to study computer science without hesitation. That advice has changed. His view is that coding is effectively a solved problem, even if the market has not fully absorbed that yet, and that a traditional CS curriculum will be less valuable than one emphasizing creativity, design, and an understanding of what problems are worth solving. He does allow that a light technical foundation remains useful for knowing where to push AI tooling, but frames that as a different course of study, not a four-year CS degree.

Browser Wars Unlikely to Crack Chrome's Dominance

On the AI browser competition, Rose is measured. He rates OpenAI's browser as snappier than expected and credits the category with finally producing genuine browser innovation after years of stagnation. But he frames the competitive dynamic through the lens of switching costs. A product that is ten percent better cannot overcome the inertia built by iMessage-style lock-in or default OS placements. The more likely outcome is a Firefox-style scenario where AI browsers reach double-digit market share among technologists without meaningfully threatening Chrome. For the companies behind them, that is acceptable because browser distribution is table stakes for broader AI market positioning, not a standalone business.

Social Media Fragmentation Favors Intimate Communities Over Scale

Rose sees Twitter alternatives as illustrative of a broader structural shift rather than a head-to-head competitive story. Bluesky sits at roughly 30 million monthly active users and appears stable rather than declining. But the more interesting thesis is that the unit of value in social media is shifting. A subreddit with a few thousand highly engaged users around a niche topic like Japanese woodworking can generate more real information density than a platform optimized for scale. Rose expects decentralized infrastructure, along the lines of Mastodon's protocol model, to serve as connective tissue between these micro-communities rather than any single platform winning outright.

He also noted a 2005 product anecdote that inadvertently documented something durable about online behavior. An early comment-voting system he built expected users to bury bad comments and move on. Instead, users expanded collapsed comments specifically to engage with them negatively before re-burying them. The dunk, as a social behavior, predates modern social media by at least two decades.