Comment by hauleth
1 day ago
I agree that any company would struggle in such case. The thing is that everyone see that GH is pushing for more agents, their Copilot thingy, and AI everywhere, while basic functionality that people relies on is constantly failing.
If you push a lot of new features but your baseline is constantly failing, then something is wrong.
If you're seriously using agents, you'll know that if they didn't offer that then people would rapidly switch platforms if they didn't. Maybe not all of them yet, but soon it will be all.
Switch platforms to what?
Switch to a git provider that offers agentic augmentation of your workflow. And I don't necessairly mean the way it works right now - it's being refined & adjusted & infrastructure is being built as we speak.
For example, in our company, most commits on main currently have 3-5 authors (we squash): 1-2 humans, 1-3 agents (cursor cloud agent getting started, ppl pulling it into cursor locally to continue, then review using copilot review, modify using copilot agent) then use a vibe coded github app offloading UI test execution to a beefy baremetal machine to adjust baselines.
Copilot review in particular is just so good, better than any agent i know (incl opus 4.7). It just allows you to skip the first few review rounds by humans and fix simple but hard to spot logical bugs, keep docstring & style up to date across the codebase, before you give it to a human - which means everyone can focus on writing more code.
Setting all of this up, at a massive scale, is just not feasible for any of these projects.