I'm in a well funded somewhat greenfield org - not a startup - we use mcp servers of that specific ticket tracking system within the coding tools to handle ticket creation, assignment, tracking, status, completion, updated by what code is being hit. mostly inheriting existing workflows like git comments, pull requests, tickets referenced in those already, just AI writes those commit messages and PR comments too. reports about all of that are made by AI as well to the stakeholder that needs to see that distilled in their language.

We won't be hiring middle management, no product managers, no engineering managers, VPs

The only aspect we don't have solved is a buffer between sales/execs and engineers, but all other functions are automated away alongside other AI assisted coding that there actually is bandwidth for the schizophrenic ideas. Things that used to be tech debt and not prioritized by engineers without management suddenly are all solved, AI makes the cleanest REST API's I've ever seen, obscure verbs properly implemented immediately. Test cases all done.

It's working really well and the friction with non technical PMs and hierarchies is gone

its a 1 liner to add relevant mcp servers to Claude Code, and every ticket tracker already has an mcp server out

for triaging between UX designers, we also just don't. we use an mcp server for UX, I can point playwright - which is usually used for testing your own site - at a competitor's website and feed all the UX information and implementation into Claude Code to promote the synthesis of an extremely advanced and already engaging design pattern into the project

at this point, I would say its a lack of competence to manage a software project or org, any other way. Amazon's deep cuts are a shot across the bow to others that know they need to "do the needful", and will be watched closely