With LLMs I've managed to ignore these issues so far on my django app with 100+ views and 50+ models. So yeah, I get that for free. My app is more complex than a simple CRUD app, but still at least 90% CRUD.
To be fair it sounds like LLMs are managing that for you. You’ve got two (mostly) black boxes stacked on top of one another here. You’ve avoided managing your DB and avoided managing your ORM. What will you do when something eventually goes wrong? Ask the LLM to fix it?
Yeah. Or apply my brain. Reviewing LLM-written code is no different to working with code from a colleague or library. It's the best of both worlds, certainly while iterating quickly on the MVP & early stages.
The benefits far outweigh the disadvantages.