There's a sentiment in the industry right now that Software as a Service (SaaS) is dead in the water. Why spend X amount of money on service providers when you can develop your own tool for a fraction of the price? With LLMs lowering the barrier to entry, the math seems obvious.
The value proposition is attractive, but it's flawed. The people that share this belief are marginalizing the importance of software developed to address the needs of many over the needs of one. A mature SaaS product isn't just code. It's a structured solution built around customers, addressing compliance and audit concerns like SOC2 and ISO27001. Scalability, reliability, and security. Years of edge cases discovered and patched. Unless you're helming your replacement with senior-level development talent iterating over it with an LLM, you're going to end up with an inferior product. And even then, you're now on the hook for maintaining it.
You might argue that's fine, and that may be true in the case of prototypes or experiments. But there's another key component to the success of a product: trust. If your application breaks down, is bug-laden, or otherwise compromised, that's a devastating blow to trust. And trust, once lost, is incredibly difficult to rebuild. Your customers don't care how you built it. They care that it works, every time.
I think the future of SaaS lies in more rapid feature development. Leveraging AI to accelerate their own roadmaps while maintaining the principles that helped build that trust and customer base. The companies that will thrive aren't the ones being replaced by vibe-coded alternatives. They're the ones using AI to ship faster, iterate smarter, and widen the gap.