New post on Polyglot AI Agents #2465
Conversation
|
🎊 PR Preview 4992700 has been successfully built and deployed to https://quarkus-site-pr-2465-preview.surge.sh
|
|
This should be what is needed to have multiple authors support. Cc. @insectengine |
When I spin up the change locally, I don't see either author. Additionally, Please don't use the same content for both title/synopsis in blog posts. It's supposed to be a very brief summary of what the readers can expect in the article. I also use this content if the blog article is included into the monthly newsletter. It's better if the author writes it verses me putting together something that could be inaccurate. |
|
Thanks for reviewing the changes!
Is missing a comma to separate authors.
Do you mean we should use a "truncate" early in the article? |
Not truncate, as that really doesn't give the reader any clue as to the intent of the article. I mean a short (sentence) that gives that helps pull the reader into the article. Examples from other blog posts... Title: Continued Focus on Native Title: How to Use Your A2A Server Agent in a Distributed or Cloud Native Environment Title: How to Secure Your A2A Server Agent with Keycloak OAuth2 |
|
Can you please update the PR so we can see if the latest changes are reflected in the preview? |
|
Thanks @insectengine for explaining, fixed in: mariofusco#2 |
|
Preview looks good now, maybe @mariofusco can add his headshot 😅 but doesn't need to be included in this PR I think 🙂 |
|
Thanks for this article! I'll say right off the bat, that I am struggling to understand the context of the post, meaning that as I have not read https://blog.mozilla.ai/wasm-agents-ai-agents-running-in-your-browser/ and https://blog.mozilla.ai/3w-for-in-browser-ai-webllm-wasm-webworkers/ (and don't really plan on doing so, and I believe most of the readers of the Quarkus blog will be in the same boat), I am scratching my head trying to understand what problem we are trying to address. IMHO, the post would benefit heavily from a brief explanation of the problem and a summary of the linked blog posts. |
|
Can we separate the blog multi-author thing from the post? I can review the post, but not the multi-author thing. |
cescoffier
left a comment
cescoffier
left a comment
There was a problem hiding this comment.
The blog conveys some great ideas, but we need to rework the form and message. Typically, I still fail to understand the one-line mission statement of the post - it talks about many things and, in my opinion, lack of clear technical insights (I want to see code :-) )
I would also avoid over-selling this as "enterprise-grade" deployment.
| --- | ||
| :imagesdir: /assets/images/posts/agentic | ||
|
|
||
| In two recent posts about WASM agents, Davide Eynard and Baris Guler show, respectively, https://blog.mozilla.ai/wasm-agents-ai-agents-running-in-your-browser/[how to bring agentic frameworks into the browser] and https://blog.mozilla.ai/3w-for-in-browser-ai-webllm-wasm-webworkers/[how to extend them] with in-browser inference and support for multiple programming languages. This article explores a different point of view: is it possible to leverage the JVM's polyglot capabilities to create a server-side equivalent, that's equally self-contained, but optimized for enterprise deployment? The result is a blueprint that showcases how the JVM can serve as a viable polyglot runtime for AI agents, combining the performance benefits of WebAssembly with the reliability and maturity of Java's ecosystem. |
There was a problem hiding this comment.
What does that mean? Do they want to run the model in the browser? Why? What's the use case?
There was a problem hiding this comment.
I'm not sure of the intro - first I would remove the first sentence, I don't understand it (see my previous comment). Then, I would tone down "enterprise deployment".
I'm also lost as you mention JVM polyglot capabilities (which I believe it's truffle, groovy...) and then you mention WASM.
|
|
||
| == Why the JVM for AI Agents? | ||
|
|
||
| The browser-based approach has its advantages, like privacy, offline capability, and user control; enterprise environments often require different characteristics: centralized management, resource optimization, security controls, and integration with existing infrastructure. |
There was a problem hiding this comment.
Well, I would say: Why the browser for AI agents?
|
|
||
| Each language brings its strengths to the AI agent ecosystem: | ||
|
|
||
| *Rust*: High-performance systems programming with memory safety, compiled to WebAssembly for maximum efficiency |
There was a problem hiding this comment.
Are you really sure that Rust in WASM is more efficient than bare metal RUST?
|
|
||
| *JavaScript*: Dynamic scripting with QuickJS integration, enabling runtime flexibility | ||
|
|
||
| The beauty of this approach is that all these languages run within the same JVM process, sharing resources and memory efficiently while maintaining their individual characteristics. This means that whatever language you used to build your agent can be integrated into this system, possibly together with other agents written in different languages. |
There was a problem hiding this comment.
One of the issues with this model is that you don't need one observability stack but 5: Java + one for each WASM language - you also need to implement tracing right, metrics...
If you want to talk about enterprise grade deployment, it should be addressed somehow.
There was a problem hiding this comment.
Oh, and multi-threading without duplicating contexts.
|
|
||
| === Try It Yourself | ||
|
|
||
| Getting started is straightforward with our Docker-based approach: |
There was a problem hiding this comment.
I need to see code to understand - right now, I don't really understand what's in WASM module, what's outside and how the different things are implemented.
|
A bit of context: This blog post will be published in early December on https://blog.mozilla.ai/ The source code is available here: https://github.com/roastedroot/wasm-java-agents-blueprint The review is fair, and the post definitely needs more context and refinement before it can be published on the Quarkus blog. I’ll coordinate with @mariofusco to make that happen. |
5 participants
/cc @andreaTP @cescoffier