Good morning everyone and happy MonDev ☕
Ready to start a new week and try something new maybe? By now you know I am always looking for new stimuli, some are just passing through, others come and go, while some arrive and decide to stay for a long time.

In the last category certainly fall the growing love for Rust and Tauri, with which I continue to build the tool with which I "deliver" this newsletter now 90%. For the curious, know that I am working on a series of articles to tell you about what I am learning with this project and share it with you! I hope to release the first part this week, so if you are interested in the topic, keep an eye on my channels from time to time 😎

But now coming to today's tool, let's talk about a topic that comes and goes, but it's always fun to play around with. I'm referring to AI; as you know, I'm not an AI-addicted person, but I like to have my share of fun with it!
And a few days ago I came across the program that I am presenting to you today which aims to be very interesting and fun!
Rivet, this is the name of the tool of this week, is a visual development environment for AI agents. But in what sense?
Well, once you download the software, you will have a drag and drop interface to create a flow of information that can interact with various LLMs, whether it's GPT, Google models, or even Open models like Ollama running locally. It can handle all the classic control flows (if, loop etc.) and can integrate JS functions directly into it or import them from the project where we decide to incorporate it.

In fact, this tool exports a yaml that is read and executed according to the rules we have defined thanks to the package @ironclad/rivet-node which will execute the programmed flow and return a readable output that can be manipulated by code.

This tool is totally free and I think the nice thing is that it is essentially a no-code program that, however, has the ability to integrate code within it and seamlessly integrate into our system.

Each project is composed of multiple graphs, each of which can handle a type of interaction; moreover, each of these graphs can be called by others or directly by code. We can therefore create a chat agent with great precision and even predict sub-agents according to our needs and implement it within our website/app with ease.

Another very interesting pro is that, our graph being a yaml file, it will be constantly versioned, with all the benefits that this entails!

I really think this could be a good tool to integrate into our developers' arsenal, what do you think?

A small concluding note: the npm package documentation is not optimal; however, the entire project is open source, so, as always, if we want, we can contribute 😉

What do you think of this tool? Do you plan to try it? In general, how often do you work with AI, if you do?

As always, if you feel like it, I'm curious to hear your thoughts!
So if you feel like it, let me know! Meanwhile, as always,
happy coding 0_1