"Congrats on 2.0 — the BEAM's supervision and fault tolerance makes a lot of sense for agent workloads. One thing we've been tackling at ARU (aru-runtime.com) is the certification layer between agent intent and execution — validating outputs against a defined contract before they pass downstream. Curious whether Jido has opinions on output validation/certification, or if that's left to the application layer?"
I haven't used Jido for anything yet, but it's one of those projects I check in on once a month or so. BEAM does seem like a perfect fit for an agent framework, but the ecosystem seeming limited has held me back from going too far down that path. Excited to see 2.0!
Just a heads up, some of your code samples seem to be having an issue with entity escaping.
Love this! The timing couldn't be more perfect. I had to write my agent framework with a mix of gen servers and Oban. It's honestly a pain to deal with. This looks like it will really remove a lot of pain for development. Thank you so much!
Very eager to read through your code! I read the first version and incorporated several of its ideas into our own internal elixir agent framework. (We make use of your ReqLLM package, thanks much for that!)
I’ve found the hardest part with agent frameworks isn’t model plumbing, it’s operational boundaries: how you isolate tools, enforce time/budget limits, and recover from partial failures when an agent call chain fans out.
BEAM’s supervision model feels like a genuinely strong fit for that, especially if each tool execution can be treated as a supervised unit with clear restart/escalation semantics. Curious whether you’ve seen teams default to many small specialized agents vs fewer general agents with stricter policies.
I just LLM-built an A2A package which is a GenServer-like abstraction. I however missed that there already was another A2A implementation for Elixir. Anyway, I decided to leave it up because the package semantics were different enough. Here it is if anyone is interested: https://github.com/actioncard/a2a-elixir
I've been following this project for several months now and Elixir/BEAM is absolutely perfect for running agents. BEAM is so incredibly lightweight; IFYKYK. Theoretically you could run 1000s of agents on a single server. I'm looking forward to seeing what people who understand this build.
The core of Jido will run on a Raspberry Pi - we've even had people look at running Agents inside the BEAM where the BEAM is deployed on bare metal (embedded)
It'd be cool to see a screenshot of what 'observer' shows as the process tree with a few agents active.
Edit: for those not familiar with the BEAM ecosystem, observer shows all the running Erlang 'processes' (internal to the VM). Here are some examples screenshots on one of the first Google hits I found:
Elixir has a LangChain implementation by the same name. And in my opinion as a user of both, the Python version and the Elixir version, the Elixir version is vastly superior and reliable too.
This agentic framework can co-exist with LangChain if that's what you're wondering.
I went down this path a bit the other night, curious what OP's answer is. My mental model was that they could be complimentary? Jido for agent lifecycle, supervision, state management, etc, LangChain for the LLM interactions, prompt chains, RAG, etc. Looks like you could do everything in Jido 2.0, but if you like/are familiar with LangChain it seems like they could work well together.
Huh... excellent timing. I am working on a project that currently is handling this with bunch od npm tasks :)(I know), but it works.
Sidian Sidekicks, Obsidian vault reviewer agents.
I think Jido will be prefect for us and will help us organize and streamline not just our agent interactions but make them more clear, what is happening and which agent is doing what.
And on top of that, I get excuse to include Elixir in this project.
Let me guess, in the next 6 months, Elixir and Erlang becoming fashionable to build AI agents and then another hype cycle of AI usage and marketing of Elixir.
What's old is now rebranded, reheated and new again.
Elixir has always been fashionable to build high performance systems in. In fact, it is more suited for AI applications than any other language or framework because of the BEAM architecture and the flexibility of the language itself. I wish more people gave it a chance. You get insane performance at your fingertips with so much scalability out of the box and your code by default is less error prone compared to dynamic languages.
Just a heads up, some of your code samples seem to be having an issue with entity escaping.
https://github.com/openai/symphony
I'm not very familiar with the space, I follow Elixir goings on more than some of the AI stuff.
It is curious... and refreshing... to see Elixir & the BEAM popping up for these sorts of orchestration type workloads.
https://web.archive.org/web/20260305161030/https://jido.run/
Congrats on the release!
I’ve found the hardest part with agent frameworks isn’t model plumbing, it’s operational boundaries: how you isolate tools, enforce time/budget limits, and recover from partial failures when an agent call chain fans out.
BEAM’s supervision model feels like a genuinely strong fit for that, especially if each tool execution can be treated as a supervised unit with clear restart/escalation semantics. Curious whether you’ve seen teams default to many small specialized agents vs fewer general agents with stricter policies.
Agree on operational boundaries - it took a long time to land where we did with the 2.0 release
Too much to say about this in a comment, but take a look at the "Concepts: Executor" section - it digs into the model here
I just LLM-built an A2A package which is a GenServer-like abstraction. I however missed that there already was another A2A implementation for Elixir. Anyway, I decided to leave it up because the package semantics were different enough. Here it is if anyone is interested: https://github.com/actioncard/a2a-elixir
The future is going to be wild
There’s a growing community showcase and I have a list of private/commercial references as well depending on your goals
Edit: for those not familiar with the BEAM ecosystem, observer shows all the running Erlang 'processes' (internal to the VM). Here are some examples screenshots on one of the first Google hits I found:
https://fly.io/docs/elixir/advanced-guides/connect-observer-...
Teaser screenshot is here: https://x.com/mikehostetler/status/2025970863237972319
Agents, when wrapped with an AgentRuntime, are typically a single GenServer process. There are some exceptions if you need a larger topology.
I was curious about the actual BEAM processes though, that you see via the observer application in Erlang/Elixir.
I used Claude to learn & refine the patterns, but it couldn’t write this level of OTP code at that time.
As models got better, I used them to find bugs and simplify - but the bones are roughly the same from that original design.
It's use-case specific though - security is a much bigger topic then just "agents in containers"
The point of Jido isn't to solve this directly - it's to give you the tools to solve it for your needs.
https://github.com/agoodway/goodwizard
(Probably complimentary but wanted to check)
https://hex.pm/packages/req_llm
ReqLLM is baked into the heart of Jido now - we don't support anything else
This agentic framework can co-exist with LangChain if that's what you're wondering.
https://github.com/brainlid/langchain
As LLM API's evolved, I needed more and built ReqLLM which is now embedded deeply into Jido.
Sidian Sidekicks, Obsidian vault reviewer agents.
I think Jido will be prefect for us and will help us organize and streamline not just our agent interactions but make them more clear, what is happening and which agent is doing what.
And on top of that, I get excuse to include Elixir in this project.
Thanks for shipping.
What's old is now rebranded, reheated and new again.