Will AI Kill SaaS Software?

Non-technical people try vibe coding something simple, are amazed that the LLM produced a working solution, and don’t have enough technical sophistication to understand what their experiment demonstrated and didn’t demonstrate.

I do think some apps are going away, though. I just started a trial of Plausible last week for privacy-first analytics on a marketing website. It didn’t do quite what I wanted, was relatively unimpressive for something that will cost $9/mo, and I eventually just used Claude Code to do exactly what I wanted for “free”. These simple utilities seem doomed to me.

But that is fundamentally different than “product-sized” SaaS solutions. I just don’t see a VP of Marketing taking a month of her actual job to vibe code a replacement for the tool her department pays $500/mo for. Hugely negative ROI compared to just buying.

This feels a lot more like the ‘90s and ‘00s when the market was trying to rationalize how open source software would impact proprietary software. Some stuff went away because OSS is good enough or better. Many proprietary software packages continued to grow. New hybrid business models emerged. And pretty much everything benefitted from open source libraries handling low level functions securely.