Google tried this and gave up, as it didn’t work out[1].
For now, there’s a lot of focus on automating software dev jobs, but I think companies are just starting to realize automation will thin the management layers as well.
I think it might actually be easier to automate middle management than it would be technical folks. It's not exactly difficult to shove an AI agent into a meeting, have it record everything, and ask everyone for updates. Likewise, it isn't hard to have an AI agent just read through tickets and prompt people for updates. Dystopian? Yeah. Hard? No. Also, the AI would be better than most middle managers I've encountered. The sad thing is, I've also had a few excellent middle managers who helped me grow as a professional.
Which is what we're seeing right now - middle management, and vestigial organizations not related to the core dev team (Trust & Safety and co.) getting trimmed.
As for LLMs, I am perfectly capable of opening a notepad, taking notes, and converting them into Jira tickets (as I have done in the past) without bleeding edge machine learning models
And in a really dystopian world, the AI could review everyone’s PRs and documentation and rank engineers based on their contributions. It could even make compensation decisions based on that. Much of the performance management can be automated (with human in the loop).
Then you could have fewer managers managing more people (and get rid of a bunch of layers of management overhead). They could focus on the higher value parts of management like coordinating between teams, making tough execution decisions, resolving technical blockers / ambiguity, etc
For now, there’s a lot of focus on automating software dev jobs, but I think companies are just starting to realize automation will thin the management layers as well.
[1] https://bpb-us-e1.wpmucdn.com/sites.psu.edu/dist/3/50375/fil...