The shift from buying software to building it yourself isn't coming—it's already here, and the companies selling you subscriptions are starting to sweat.
The Summary
- Thomas Paul Mann, CEO of Glaze (and former Raycast CEO), predicts half the software you use in a few years will be tools you built yourself — not downloaded from an app store
- Glaze is building AI-powered software that lets anyone create custom apps through natural language, no coding required
- The "SaaS apocalypse" thesis: when users can generate exactly what they need on-demand, why pay monthly subscriptions for bloated, one-size-fits-all tools?
- Mann calls this "disposable software" — apps you spin up for a task, use once or twice, then forget about
The Signal
Mann isn't selling vaporware. Raycast already proved he knows how to build tools power users actually adopt. Now he's making a bigger bet: that the entire software distribution model is about to flip inside out.
The core insight is simple but radical. You don't need a team of engineers to build most software anymore. You need a good prompt and an AI that understands what you're trying to do. Glaze aims to be that AI, letting users describe an app in plain English and getting back working software in minutes.
"Why pay for 100 features when you only need three, and you could build those three yourself in an afternoon?"
This isn't about replacing Figma or Excel. It's about all the micro-tools that live in the long tail. The custom dashboards. The internal workflows. The niche calculators and trackers that would never justify a $20/month SaaS subscription but would be worth 20 minutes of your time to build once.
Mann calls these "disposable apps," but that undersells what's happening. The real shift is from software-as-product to software-as-artifact. You don't buy it. You don't subscribe to it. You generate it when you need it, like asking ChatGPT to write an email. Except instead of text, you get a functional app.
The economics break down fast for traditional SaaS companies when this becomes real:
- Customer acquisition costs stay the same or rise
- Retention drops because users churn to their own tools
- Feature bloat becomes a liability, not an asset
- The moat shrinks to data and integrations, not functionality
Mann points to OpenAI's Canvas and Anthropic's Artifacts as early proof of concept — users already creating throwaway tools inside chat interfaces. Glaze is betting the next step is making those tools persistent, shareable, and actually useful beyond a single session.
The counterargument is that most people don't want to build their own tools. They want someone else to handle it. That's probably still true for 80% of users. But the other 20% — the power users, the tinkerers, the people who used to learn Excel macros or write Python scripts — that's where the real economic damage happens. These are the users who pay for premium tiers, evangelize products, and build internal momentum at companies. When they leave, the bottom falls out.
The Implication
If Mann is right, the next decade of software looks less like the App Store and more like Legos. You'll still use Gmail and Slack and Notion. But the connective tissue between them — the custom workflows, the reporting dashboards, the weird little tools that make your job easier — you'll build those yourself.
For founders, this means rethinking what a defensible product looks like. For workers, it means learning to think like a builder, even if you never write a line of code. The companies that win in this world won't be the ones with the most features. They'll be the ones with the best APIs and the cleanest data to plug into whatever users decide to build next.