
Google scaled the monorepo. Meta did too. Anthropic just pushed it further for AI - but is the next answer really a bigger repo? But AI changes the bottleneck. Code is becoming abundant. Verification, ownership, and coordination are the hard parts now. A repo tells you what code changed. A graph tells humans and agents what that change means — who owns it, what it affects, and what needs to be verified. So anyone can make a change without understanding the entire system — or being afraid of what they might break. That’s why the next abstraction isn’t a bigger repo. It’s product units connected on a graph. bit.cloud/blog/a-bigger-repo…







