AI Agentic Frameworks are building their own Erlang/Elixir

AI Agentic Frameworks are building their own Erlang/Elixir
Photo by Ashkan Forouzani / Unsplash

Looking at what some frameworks are doing, I feel they are replicating in Python what we already have in the BEAM.

They talk about orchestrating small agents that communicate through messages (Langroid actually mentions being inspired by the Actor model).

There are hints of transforming texts and composing functionality. Very functional-esque.

Some mention error recovery when executing agent tools. Well, BEAM has supervision trees and fault tolerance built into the language itself.

Sometimes state is not shared globally (a relief!) and it is propagated through messages as well. That's precisely how immutability and BEAM processes work.

What I think is missing in this ecosystem is a good way of exchanging messages (we have that in the BEAM, and a message hub is easy to use) and testability (I don't see these frameworks focusing so much on how to unit-test such complex behaviors).

That's why I'm excited to see the focus on AI and Agents on Elixir that Mark Ericksen, Mike Hostetler, Thomas Miller and others (even myself!) are working on.

Elixir (and the BEAM) is the perfect ecosystem for the new Agentic way of building software.


Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to George Guimarães..

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.