Muhammad Ibtisam

Posted on Apr 19

Using CrewAI with Ruby Without the Boilerplate

#ai #ruby #crewai #gap

The Ruby ecosystem has always been great for building clean, maintainable, production-ready systems.

At the same time, tools like CrewAI are opening up a new world — multi-agent workflows that can research, generate, and collaborate on their own.

Naturally, I wanted to try CrewAI from Ruby.

But honestly… the experience wasn’t what I expected.

⚠️ The Gap

Using CrewAI in Ruby usually meant going a bit too low-level:

Writing raw HTTP requests
Manually handling authentication
Building your own polling for async runs
Dealing with inconsistent error handling

Sure, it worked.

But it didn’t feel like Ruby.

And if you’ve worked with Ruby for a while, you know that feeling matters.

We care about clean APIs, readability, and developer happiness.
Integrations should match that vibe.

💡 The Idea

I started thinking:

What if interacting with CrewAI felt just like calling a simple Ruby service object?

Something that:

Removes boilerplate
Feels predictable
Fits naturally into any Rails or Ruby app

So instead of scattering HTTP calls everywhere, I built a small client to wrap it all cleanly.

🧩 Introducing: ruby-crewai

ruby-crewai is a lightweight Ruby client for the CrewAI AMP API.

The goal is simple: make working with AI crews feel native to Ruby.

✨ Example

client = CrewAI::Client.new(
  access_token: ENV["CREWAI_TOKEN"],
  uri_base:     ENV["CREWAI_URI_BASE"]
)

client.kickoff(inputs: {
  topic: "AI in healthcare",
  audience: "CTOs"
})

And… that’s pretty much it.

No manual requests.
No custom polling logic.
No unnecessary complexity.

⚙️ What You Can Do

With ruby-crewai, you can:

Kick off AI crews
Check execution status
Resume runs with human feedback
Handle errors in a clean, structured way

All through a simple, Ruby-friendly interface.

🧠 Why This Matters

Using modern AI tools shouldn’t mean stepping outside your ecosystem.

By making CrewAI feel natural in Ruby, you get:

Faster prototyping
Cleaner integrations
Less “glue code”

Which basically means:

👉 More time building actual features, not wiring things together

🚀 Getting Started

Install the gem:

gem install ruby-crewai

Check out the project:

👉 https://github.com/MuhammadIbtisam/ruby-crewai

🔮 What’s Next?

This is just a small step toward better AI tooling in Ruby.

There’s still a lot to explore:

Multi-agent workflows
Human-in-the-loop systems
AI-powered backend services

If you’re experimenting with CrewAI or building AI features in Ruby, I’d genuinely love to hear how you’re approaching it.

❤️ Final Thoughts

Ruby has always been about developer experience.

AI integrations should feel the same — simple, expressive, and actually enjoyable to use.

DEV Community