To me, the biggest hurdle to trying to manage agents like humans is that there's no real continuity, out of the box at least.
You can trying to get around this with RAG and markdown files and skills but you're basically building from scratch the "tools" on how to remember the codebase that you take for granted with people
There's no continuity with outsourced consultants either. But still people manage to make it work.
With the exact same tools you use with LLMs: proper documentation and detailed instructions.
Then you can grab any random consultant or LLM with a fresh context and get them to work.