Better AI models have changed how I use software. The interesting part is not the tools themselves, but the decision logic behind them.
Features I used to pay for are now solved by general-purpose models. Grammar fixes, image editing, research, and simple analytics can often be done faster and cheaper without dedicated SaaS tools.
This matters because most SaaS products still sell features, not outcomes. When a feature can be replicated with an LLM in hours, its standalone value collapses, especially during renewal cycles.
That is when finance asks a simple question: what business outcome justifies this subscription?
I see organizations moving back toward a build bias, but in a different form. Not large internal platforms, but small, disposable AI apps created for a specific need, used once, and then retired.
This shift changes SaaS economics. Products built on low differentiation will be questioned first, price pressure will increase, and renewals will become harder to defend.
Some categories will remain safer for longer. Highly regulated domains, products with strong network effects, and platforms tied to high SLA and operational risk still deliver defensible value.
For years, buy vs build clearly favored buy. AI is quietly shifting that balance back.
If you look at your current SaaS stack, which tools would survive that question today?
Leave A Comment