I understand the ecosystem argument. Once you've got your repos, CI/CD, PRs, code reviews, packages, integrations, etc. all wired into GitHub, moving everything is a pain. But Git itself is distributed, and GitHub has a pretty open plugin/integration system. There are also perfectly capable alternatives like GitLab, Bitbucket, Forgejo, Gitea, SourceHut, etc.
So what is the actual thing that makes a 10–20 person software company stick with GitHub? Is it mostly switching costs? The ecosystem? Developer familiarity? Something else? And if you were to start a project today, would you still go with GitHub?
And a slightly more personal question.
I've had a rough idea for what a new GitHub could look like if it were built specifically for software that is increasingly being written by AI agents, and specifically for companies rather than open-source projects. The idea isn't really "another GitHub clone". More of a forge designed around how software teams might work when agents are doing a significant part of the coding, reviewing, testing and deployment work, etc.
I've been sitting on the idea for a while because, well, GitHub exists. But after the recent outages, I've started wondering whether this is actually worth pursuing. In fact, I see a future where there are multiple alternatives to GitHub with their own take on the interface and UI/UX. Interested to hear your thoughts.
With the advent of AI, people are probably thinking its easy to clone services like this and many others. Sure a 1.0 might be easy to stand up but you've got to now patch it, maintain it, extend it, integrate more tools into it, and keep it running better than github. People may joke/sneer at uptime metrics but when internal tools at a software company are down or worse, buggy and causing issues with delivery, its more painful than checking a status page to see when its back.