GitHub recently introduced GitHub Copilot as a reviewer tool for pull requests. I gave it a shot with low expectations and had a pleasant surprise. The reviews you get could easily be mistaken for those from a human reviewer.

This made me think about the role that AI could play in reducing the cost of maintaining OSS. As you might know, the classic challenge of open source is the cost of maintenance. This is, in fact, the reason why we started working on evolving Tuist into a business.

Issues and pull requests pile up, you don't have the time to review them all, and you have to process the emotions that come with feeling like you're lagging behind or not meeting the community's expectations. But if most of this can be delegated to AI, that changes everything.

I believe we, open source maintainers, should invest in having the right CI checks to ensure the code that is merged is of the quality we want to ship, along with rules to guide GitHub Copilot to do its best work. How those rules are provided to AI agents is still an evolving territory, but once it settles down, I think it makes sense to invest in providing agents with that context.

Who knows... maybe we'll get to a point where GitHub Copilot can also reject PRs that don't align with the direction/vision of the project, or merge others automatically. I think that would be fantastic.

In the context of open source companies, a subject I've been thinking about a lot lately, this approach gives you access to a developer workforce that's not limited by your location or your number of employees. It means you can move much faster than your competitors, at a much lower cost, allowing you to focus on building revenue from large enterprises while innovating and watching your community foundation grow.

I don't know about you, but I think this is a unique strength. When I think of companies working in the same space as Tuist, none are approaching business this way, so I believe we're in a good position to take advantage of this opportunity.