KaibanJS - Open-source framework for building multi-agent AI systems - Interview with Dariel Vila

Since the launch of ChatGPT, there has been a lot of interest in AI systems. The question is, how do you build your agents, for example?

In this interview, we will look into a JavaScript-based solution called KaibanJS↗ with its author Dariel Vila↗.

Can you tell me a bit about yourself?

#

I’m Dariel Vila, Lead Developer at KaibanJS and a passionate advocate for JavaScript. With over 15 years in the tech industry, I’ve worn many hats — Developer, Team Lead, and Entrepreneur.

You know, JavaScript has been a game-changer for me. It’s not just a language but my ticket to a whole new life. It helped me leave Cuba, land a job in Venezuela when I had no papers, and even get my first gig in the US. Thanks to JS, I created my first open-source project, AniJS, and later founded my company, FRONT10. It’s pretty wild how a programming language can shape your entire journey, right.

These experiences have shaped my perspective and fueled my passion for the JavaScript ecosystem. They’ve also inspired me to give back to the community through projects like KaibanJS, aiming to empower other developers to achieve their goals and push the boundaries of what’s possible with JavaScript.

How would you describe KaibanJS to someone who has never heard of it?

#

Imagine you’re a project manager, but instead of managing a team of people, you’re orchestrating a group of AI agents using a Kanban Board. If you’ve used Trello or Notion, you know how that works. That’s what KaibanJS allows you to do. The JavaScript-native framework lets you build and manage multi-agent AI systems following the Kanban methodology.

How does KaibanJS work?

#

You start by creating your AI agents, each with specific roles and capabilities. These could be agents for data analysis, content creation, customer service – whatever your project needs. Then, just like in a regular sprint planning session, you assign tasks to these agents.

The magic happens on what we call the Kaiban Board. Think of it as a Trello or Notion board for AI tasks. You’ve got your columns – maybe “To Do,” “In Progress,” and “Done” – and you can visualize the status of each task in real time. You can drag and drop tasks, set priorities, and watch your AI agents work through their assignments.

What’s cool is that you can see the interactions between agents, track their progress, and even step in to make adjustments if needed. It’s like having an agile AI team at your fingertips, all managed through a familiar visual interface.

It is important to note that people can easily build their own UIs using the state management system that KaibanJS provides.

So, in essence, KaibanJS bridges the gap between JavaScript developers and the world of AI, allowing you to create, manage, and visualize complex AI workflows as quickly as you’d manage a sprint in your regular development process.

How does KaibanJS differ from other solutions?

#

KaibanJS stands out because it’s designed exclusively for JavaScript, unlike most AI frameworks that cater primarily to Python. When you deeply specialize in technology, you understand how to leverage its unique features—such as reactive state management and component-based architecture—providing tools that ensure robust performance and extensive customization. KaibanJS allows developers to build, visualize, and deploy AI workloads effortlessly within a familiar environment. The Kaiban Board is a game-changer in how we interact with and manage AI workflows, making AI development as intuitive as organizing a kanban board for sprint planning.

Why did you develop KaibanJS?

#

The inspiration for KaibanJS arose from frustration with the lack of proper AI tools for the JavaScript community. Having been deeply involved in JavaScript for years, I felt creating a framework enabling developers to harness AI capabilities without the steep learning curve associated with other languages was crucial. I wanted to bridge the gap and make AI accessible to JavaScript developers without requiring them to learn a new language or ecosystem. Additionally, improving user interfaces in AI interactions is vital, and KaibanJS, together with the Kaiban Board, excels.

What next?

#

Looking ahead, we’re excited to gather user feedback as we finalize the framework. Our next steps include:

1. Building a solid community around KaibanJS
2. Keep improving the framework with capabilities that JavaScript developers need in real-world applications.
3. Try to push the boundaries of what’s possible with AI in JavaScript.

What does the future look like for KaibanJS and web development in general? Can you see any particular trends?

#

The future looks incredibly promising! With the increasing demand for AI capabilities across various platforms, KaibanJS is positioning itself at the forefront of this evolution. I envision a trend where AI integration becomes as common in JavaScript frameworks as using a state management library is today.

One exciting development is the emergence of smaller, more efficient language models that can run directly in the browser or on devices. This shift is a game-changer for JavaScript developers because it opens up new possibilities for AI-powered applications without constant server communication. Imagine having powerful AI capabilities in your Chrome browser or smartphone, all powered by JavaScript!

JavaScript’s ubiquity is a huge advantage here. It’s already everywhere - in browsers, servers, mobile apps, and IoT devices. This versatility means that as AI becomes more integrated into our daily tech, JavaScript developers will be uniquely positioned to create AI-enhanced experiences across various platforms.

What advice would you give to programmers getting into AI?

#

My advice is simple: dive in! It’s easier than you might think. AI development primarily involves consuming APIs and understanding how to interact with them. As with ReactJS, you don’t need to understand the framework’s internals to build software effectively.

Many beginners mistakenly believe they must dive deeply into concepts like Neural Networks. Instead, I encourage you to:

1. Grab an LLM API and start experimenting
2. Create something tangible, like a simple chatbot or content generator
3. Use tools like KaibanJS to make the process more engaging and productive
4. Focus on leveraging AI to enhance your JavaScript projects
5. Don’t be afraid to make mistakes – they’re part of the learning process

Remember, the goal isn’t to become an AI researcher overnight but to explore how AI can enhance your development skills and projects.

Who should I interview next?

#

I highly recommend interviewing Zeno Rocha, the founder of Resend. I’ve been following his work since I was 26 years old, and he has inspired many frontend developers.

Any last remarks?

#

I’m genuinely excited about the possibility of every JavaScript developer becoming an AI developer, regardless of their prior experience with machine learning or data science.

By leveraging the familiar JavaScript environment and providing intuitive tools, we can lower the barrier to entry for AI development.

We’re facilitating innovation in this domain and fostering a vibrant community where JavaScript developers can support one another in their AI journey.

With KaibanJS, we’re democratizing AI development and opening up a world of possibilities for JavaScript enthusiasts everywhere. We aim to make working with AI as natural for JavaScript developers as building a web application – no specialized knowledge is required.

Conclusion

#

Thanks for the interview, Dariel! It is exciting that you mixed AI with JavaScript in such a creative way. I have been looking into agent development lately, and KaibanJS fits the bill nicely!

To get started with your own agent adventures, you can learn more about KaibanJS online↗.