October 15, 2026
Today is the final day of TechCrunch Disrupt 2026 in San Francisco, and according to TechCrunch's own rundown of the closing lineup, the two sessions worth paying attention to aren't really about different topics. They're about the same topic, told from opposite ends of the transaction. On the Disrupt Stage, Amjad Masad, CEO and co-founder of Replit, is doing a conversation billed as 'Everyone Can Build Software Now. What Happens Next?' Per the official Disrupt agenda, the premise is that AI is letting millions of people create software for the first time, and Replit is positioned at the center of that shift -- a conversation about the future of programming, entrepreneurship, and what happens when anyone can turn an idea into a product. Over on the AI Stage, Cathy Gao of Sapphire Ventures, Aaron Jacobson of NEA, and Barr Moses, CEO and co-founder of Monte Carlo, are running a panel called 'Rewriting SaaS: Why AI Breaks the Old Business Model,' built around whether the SaaS playbook is dead or evolving, how AI products should even be priced, and how vendors build any kind of defensible moat once the underlying models are commodities anyone can call.
Both panels are circling the same fact: AI has pushed the cost of producing working software toward zero. Neither one disputes that. Where they split is over who that fact is actually about.
Masad's framing is, at its core, about headcount -- specifically, turning people who were never developers into people who write prompts, review AI-generated code, and are on the hook for what ships. That's a real and valuable thing. It's genuinely exciting for a solo founder, a student, or a hobbyist who wants to build something and has never touched a terminal. Replit has built an actual business on exactly that promise, and credit where it's due: the tooling to go from idea to a running app has never been this accessible.
But notice what the framing quietly hands off to the business. If you are the person (or the two-person team) who vibe-coded your internal tool or customer portal into existence, you are now also the person responsible for hosting it, patching it when a dependency has a security hole, and understanding it well enough to change it when the business needs change -- which it always does. The pitch is 'anyone can build.' It doesn't say much about who's still around in six months to explain why the code works the way it does, or what happens when the one person who built it leaves and the AI that generated it has no memory of the decisions it made. That's not a hypothetical edge case. It's the normal lifecycle of every piece of software a growing business depends on.
The SaaS panel is asking a more commercially honest question, and it's one closer to what an actual business owner is wrestling with. The question was never really 'can AI write this code' -- that part is settled, demonstrably, by every demo at this conference. The live question is: should a business keep paying recurring per-seat fees, indefinitely, for software it uses a fraction of, now that building a replacement is cheap? Gao, Jacobson, and Moses are debating this from the investor side -- what happens to moats, pricing, and the SaaS playbook when the cost of the underlying product collapses. But the same question, pointed at a business's own P&L, is the one worth answering this quarter: which line items on your software budget are paying for capability you actually need, versus paying for a vendor relationship that exists mostly out of inertia? That's the exact calculation we walk through in Buy vs. Build vs. ViibeStack, and it's a very different question from 'can I build this.'
Here's our view: the distinction between these two sessions isn't about ambition or audience size, it's about liability. Masad's framing bets that more people becoming developers is the future -- and for builders and tinkerers, it is. But most businesses don't want to become developers, including after launch. They want the thing to keep working without someone on staff fluent in whatever the AI generated. The real dividing line in this whole AI-and-software shift is: who's on the hook when the code needs to change six months from now -- the business itself, or the platform that built it?
That's the bet ViibeStack is actually making, and it's why our AI App Builder isn't trying to turn your office manager into a software engineer. It's trying to make sure that when your pricing changes, your workflow changes, or the person who set up your CRM moves to a different role, the system keeps running and keeps getting maintained -- because that responsibility sits with the platform, not with whoever happened to type the original prompt. If you've ever compared what it actually takes to keep a Bolt- or Lovable-style build alive past day one, our breakdowns on ViibeStack vs. Bolt.new and ViibeStack vs. Lovable walk through exactly where that ownership question shows up in practice.
If you're a developer or a founder who wants to build for the sake of building, Masad's session is the more inspiring watch, and Replit is a legitimate tool for that job. But if you're a business owner staring at a renewal invoice for software three of your twelve employees actually use, the SaaS panel is asking your actual question, and the honest answer isn't 'go learn to prompt an app into existence' -- it's 'stop paying for capability you could own outright, through a system that stays maintained without turning you into IT.' That's the whole case for replacing your stack piece by piece instead of either renewing blindly or taking on a pile of self-built tools nobody else on the team can touch. The cost of writing code is heading to zero. The cost of someone owning that code for the next five years is not, and that's the number worth arguing about this quarter.
Sources