No Vibe Coding Is Not Killing B2b Saas Michael Aquilina

Cursor's CEO says vibe-coded software has shaky foundations. He is right. But the real casualty is not engineering talent. It is the operational SaaS vendors those teams relied on. Cursor is worth $29.3 billion. Its AI code editor is used by more than half the Fortune 500. And its CEO, Michael Truell, just told Fortune that vibe coding builds "shaky foundations" that eventually "start to crumble." He is not wrong. But he is warning the wrong people. The most prolific vibe coder in tech right now is not a developer.

Jason Lemkin, the founder of SaaStr, has shipped more than 10 production applications since summer 2025 using Replit and Claude Code. Those apps have been used close to a million times. His startup valuation tool, SaaStr.ai, hit 500,000 users in its first 45 days. Subscribe to our newsletter and never miss a story. No spam, ever. None of those tools replaced a software engineering team. They replaced the SaaS products those teams used to do their jobs. Marketing automation platforms, customer success software, internal reporting dashboards.

The vendors who sold those tools are the ones with a problem. The term was coined by Andrej Karpathy, the former Tesla AI director and OpenAI co-founder, in a February 2025 post on X. "There's a new kind of coding I call 'vibe coding,'" he wrote, "where you fully give in to the vibes, embrace exponentials, and forget that the code even exists." The phrase stuck. Merriam-Webster added it within weeks. Collins Dictionary named it Word of the Year for 2025.

Monthly Google searches for "vibe coding" now exceed 110,000 in the United States alone. In practice, vibe coding means describing what you want a piece of software to do in plain language, then letting an AI model write the actual code. Tools like Cursor, Replit, GitHub Copilot and Claude Code handle the implementation. The human steers. The machine builds. While Lemkin was shipping apps, his Chief AI Officer at SaaStr, Amelia, was building something more consequential. She vibe coded two internal tools that now run core business functions.

The first, called 10K, handles almost all of SaaStr's marketing operations. Not assists them. Runs them. Whatever combination of HubSpot, Mailchimp, analytics dashboards and scheduling tools the marketing team previously relied on, 10K replaced. The company went from a full marketing stack to a single custom-built tool. The second, called Qbee, manages the operational relationship with sponsors across what is now a $10 million-plus sponsorship business. Contracts, deliverables, communications, tracking.

That is the kind of workflow that used to require a CRM subscription, a project management tool and probably a handful of Zapier integrations gluing them together. Now it is one vibe-coded application. Both would have taken a traditional development team months and real budget to build. Amelia shipped them while Lemkin was on what he calls his "vibe coding sabbatical." "Vibe coding scales when more than one person on your team can do it," Lemkin wrote. "We have two builders now instead of one.

That matters enormously." Truell's concern about shaky foundations is legitimate for production software that needs to scale, handle edge cases and survive security audits. A vibe-coded SaaS product competing with Salesforce's engineering team will hit structural limits. But a vibe-coded internal tool does not need to compete with Salesforce's engineering. It needs to replace Salesforce's invoice. If the tool breaks, the person who built it can describe the fix in plain language and have it patched in minutes. The per-seat licensing fees disappear either way.

The AI coding tools market is worth $7.37 billion and growing. Around 85 percent of developers now use at least one AI coding tool in their workflow. GitHub Copilot alone has 4.7 million paid subscribers. Cursor's own user base spans more than half the Fortune 500. But the signal from SaaStr suggests the bigger addressable market is not developers writing better code faster. It is non-developers building bespoke tools that cancel SaaS subscriptions. Lemkin stopped vibe coding for 90 days. Not because the tools got worse.

Because the process nearly burned him out. "Vibe coding is not passive," he wrote. "It's relentless iteration, debugging sessions that spiral, product decisions made at 11pm, and a constant low-grade anxiety about whether the thing you shipped actually works." Two hours a day, every day, building production software while running an eight-figure events business, managing a venture fund and trying to maintain a personal life. Lemkin describes vibe coding as a "sprint capability" for non-developers. You can go deep for weeks, ship something meaningful and then you need to recover.

Professional developers have 10,000 hours of pattern recognition that lowers the cognitive load. Everyone else pays the full energy cost on every build. The feedback loop is addictive. You ship a product, real people use it, and you want to build the next thing immediately. But that cycle is not sustainable at full intensity. Business Insider just launched a dedicated newsletter called "Vibe Mode." The Wall Street Journal, The Times and Forbes are all running regular coverage.

When a coding technique gets its own editorial beat, it has crossed from niche curiosity to mainstream business tool. The pattern emerging is not that vibe coding will replace software engineering. Truell is right that the code quality is not there for that. The pattern is that vibe coding will gut the operational SaaS market. Marketing automation, customer success platforms, internal reporting tools, workflow orchestration. The entire category of software that existed because building custom alternatives was too expensive. That cost just collapsed.

The disruption will not stop at the application layer. Platforms like Vercel and Netlify built their businesses on being the simplest way for non-developers to deploy web applications. That positioning felt durable six months ago. It is less durable when the same AI agent that helped someone build the app can walk them through deploying it on a $5 VPS using open source tools. Docker, Coolify, Caddy, Nginx. The ecosystem is mature, well documented and free.

The only thing that kept non-technical users away from it was the learning curve, and that is exactly what vibe coding removes. The only winners left standing are the AI companies themselves. Anthropic, OpenAI and Google are already building developer ecosystems designed to lock in exactly this kind of usage. SaaStr is the proof of concept. A non-technical founder and his Chief AI Officer replaced multiple paid software platforms with vibe-coded tools in under a year. The tools ran. Users came. The SaaS subscriptions did not get renewed.

Lemkin himself has written about what he calls the "90/10 rule" for replacing paid SaaS with vibe-coded apps. If the tool covers 90 percent of what you need and costs nothing in licensing, the missing 10 percent is a rounding error. The irony is that vibe coding may end up threatening developer jobs after all, just not in the way anyone expected. It will not be AI writing better code than humans. It will be AI destroying the revenue of the SaaS companies that employ those humans.

Most software developers work for companies that sell operational software to other businesses. If those businesses stop buying and start building, those companies shrink. Andrej Karpathy, the former Tesla AI director and OpenAI co-founder, introduced the term in a February 2025 post on X. He described it as a style of programming where you "fully give in to the vibes, embrace exponentials, and forget that the code even exists." The phrase was added to Merriam-Webster within weeks and named Collins Dictionary's Word of the Year for 2025.

Yes, though with limitations. Non-developers have used tools like Replit and Claude Code to ship production applications that handle real business operations, from marketing automation to customer relationship management. Some have reached hundreds of thousands of users. However, Cursor's CEO has warned that vibe-coded software builds "shaky foundations" and may not hold up for complex, production-grade engineering projects. The sweet spot so far is internal tools and operational software where reliability matters more than architectural elegance.

Current evidence suggests it will not replace developers but will displace the operational SaaS tools those developers built. The AI coding tools market is worth $7.37 billion, and 85 percent of developers already use AI coding assistants. But the clearest disruption so far has been in marketing automation, customer success platforms and internal tooling, where non-developers are building custom software that replaces paid SaaS subscriptions. The vendors charging per seat for operational workflows are more exposed than the engineers writing code. Not at high intensity, according to practitioners.

Non-developers who have shipped production apps describe the process as cognitively exhausting, with debugging sessions that spiral and a constant low-grade anxiety about whether what they shipped actually works. Some treat it as a "sprint capability" rather than a daily habit, going deep for weeks to ship something meaningful and then stepping away to recover. Professional developers have thousands of hours of pattern recognition that reduces the mental load. Everyone else pays the full cognitive cost on every session.

London Tech Week returns to London Olympia from 8 to 12 June with a new Deep Tech Stage spanning quantum computing, space, surgical robotics and life sciences. London and San Francisco-based Zalos has raised $3.6M in seed funding to build autonomous AI agents that log into existing accounting and finance platforms and do the repetitive work humans have been stuck with for years. Coral Springs attorney Matthew Fornaro has spent 23 years advising small businesses, but his most effective legal work happens in a classroom.

After collapsing under the weight of his own success, former CEO Julien Godbarge reverse-engineered his burnout into 10 leadership blind spots and wrote a business fable to help other executives see them coming.

People Also Asked

• No, vibe coding is not killing B2B SaaS - Michael Aquilina
• Why Vibe Coding Spells Trouble for SaaS - Built In
• Vibe Coding Is Replacing SaaS Tools, Not Software Developers
• "Vibe Coding" is Awesome. But No, It's Not the End of SaaS. | Cyferd
• Why Vibe Coding Won't Kill SaaS Businesses - startuphakk.com
• Why Vibe Coding Won't End SaaS: A New Perspective
• Why Vibe Coding Won't Kill SaaS (And When to Build Anyway)