I'm a little surprised by the sentiment here that LLMs don't do well with Elixir. I've had a pretty good experience using AI tools on Phoenix/Elixir side projects.
I've only used LLMs with Elixir, so I don't have any other experience for comparison, but I've found that although Claude frequently employs the wrong approaches in Elixir, I usually know when he'll have trouble and just ask him to read pertinent documentation first. So long as he's read the manual he seems to do just fine.