If software becomes ephemeral

Every month there's another post declaring SaaS is dying. The argument goes something like this: eventually we won't pay for much software at all. Instead, we'll just ask AI to generate custom software and it will create it instantly.
We're not there yet. It still requires a lot of expertise to create software. But eventually, it's hard to see how ephemeral software will not become a reality.
What are the implications of ephemeral software? Will it truly kill off the software industry?
My own view is that we'll still purchase software. Sure we'll probably all use AI to generate some things, but we'll also appreciate people who build software for us. Why? Because at this stage custom software will become more like art.
What makes a painting valuable? It's not the canvas. It's not the paint. A Picasso is valuable because it's a Picasso — a work that carries a particular style, a particular point of view, created by a specific person with a specific body of work. The raw materials are commodity. The vision is what you're paying for.
Software is heading the same way. When anyone can generate an app from a prompt, the ability to create software stops being the differentiator. Instead, the question becomes: is this the kind of software you actually enjoy using? Does it feel considered? Does it reflect a clear point of view?
Apps built by non-experts using AI tools will inevitably take on the flavor of those tools — an Anthropic smell, a Lovable smell, an OpenAI smell. Ideas baked in someone else's aesthetic. For a lot of use cases, that's perfectly fine. But for others, people will crave software with a specific identity.
To me the trajectory is clear. In a future where software can be created almost instantly, the real differentiator will be flavor. Quality, style, and consistency. Do you trust this company or person? Do you like the way their products feel? Brand has always mattered in software, but it's about to matter a lot more. Product-led growth, design sensibility, and the relationships you build with customers will outweigh the act of building the software itself.
Software may become ephemeral. But the people who make it won't be.
Comments
You might also like…
Create a daily brief with Claude scheduled tasks and ContextStore
I built an autonomous research assistant using Claude scheduled tasks and ContextStore. It scans my projects, picks topics itself, and delivers a daily brief every morning. Karpathy recently posted about a similar workflow in Obsidian — nice validation that this pattern is taking hold.

Generating ASCII wireframes and flowcharts with Claude
ASCII wireframes and flowcharts are one of the most practical things you can ask Claude to generate. I use them for UI tasks, dev plans, and increasingly as the design artifact itself.

Three areas of AI innovation
AI is advancing on three areas at once. Better models, better harnesses, and better hardware. Understanding how they work together helps you make sense of where things are headed.

Introducing the ContextStore CLI
Your AI agent can read local files, but it doesn't know about your other ContextStore spaces. The cstore CLI bridges that gap — giving any agent access to all your context from any project.
