things to try:
- slow chat
- with tools
- streaming chat
- with tools
video ideas:
- ai agents elixir conf
- recommends oban instead of GenServer https://hexdocs.pm/oban/Oban.html
- can autoscale with flame https://hexdocs.pm/flame/FLAME.html
- tool calling scaling?
- using elixir langchain for tool definitions
real time blog: https://seanmoriarity.com/2024/02/25/implementing-natural-conversational-agents-with-elixir/
elixir clustering examples and architecture: https://oneuptime.com/blog/post/2026-01-26-elixir-distributed-systems/view
process groups for distributing workers across cluster: https://memo.d.foundation/topics/elixir/pg-in-elixir