Published ·2 min read

AI Makes Open Source More Important, Not Less

AI makes custom software easier, but reliable open-source building blocks become more valuable as teams build more ambitious systems.

AI has made it easier than ever to produce code.

In theory, we should all be writing our own perfectly crafted libraries, custom-made for every use case. But that isn’t what happens. We still use the same libraries as before, along with newer ones that have proven useful.

I think that’s a good thing.

The SaaSpocalypse argument says that the future of SaaS products is uncertain because teams can now vibe code their own tools. Why spend thousands per year when you can spend one week and a modest amount on tokens building a tool that fits your organization perfectly?

That same argument could be applied to libraries. Why use someone else’s front-end state management library when you can make your own?

Because most software does not need to be custom.

I probably don’t get much benefit from making my own state management library unless I’m operating at a scale or level of specificity that existing tools don’t handle. For most projects, one of the existing solutions will be fine.

That’s why I think open source becomes more important, not less.

As software gets faster to make, we become more ambitious about what we want to build. Ambitious projects need reliable building blocks. Not everything should be invented from scratch every time. That time is better spent defining and building the features that make your software unique.

LiveKit is a good example. It gives developers open-source components for apps that need multi-user conferencing. Some components are primitive, while others are more opinionated. If your use case is standard, the opinionated components save time. If your use case is more specific, the primitives give you room to build your own thing.

That balance feels like the future to me.

Some people will stop paying for SaaS products because they can make their own version. Similarly, some builders will make their own libraries for fun, or because existing tools genuinely do not work for them.

But for the majority of software projects, the right move will still be to rely on well-made open-source software. AI does not remove the need for shared foundations. It makes those foundations more valuable.

Working through something similar?

Bring the constraint and we’ll figure out whether I can help move it forward.