Why Korben?

We built Korben because existing automation tools are either too simple (lacking AI capabilities) or too complex (enterprise-grade with steep learning curves). We wanted something in between: powerful AI agents with a clean, composable architecture for personal use.

The Problem

Modern automation should leverage LLMs, but most tools make this harder than it needs to be:

• Enterprise platforms are overengineered for personal projects
• No-code tools lack flexibility and can’t handle complex logic
• DIY scripts become unmaintainable spaghetti code
• AI libraries require too much boilerplate for simple tasks

The Korben Approach

Korben gives you the best of both worlds:

1. Simple Interface: pdm run python3 ./korben.py --task entropy
2. Powerful AI: Built on ControlFlow for sophisticated agent orchestration
3. Composability: Mix and match tasks to build workflows
4. Production-Ready: Deploy to Prefect Cloud when you’re ready to scale

Example: Entropy Task

Here’s a simple Korben task that uses multiple AI agents:

import controlflow as cf

def run(**kwargs):
    """Run the entropy task with scientist and poet agents."""
    scientist = cf.Agent(name="Scientist", instructions="Explain scientific concepts.")
    poet = cf.Agent(name="Poet", instructions="Write poetic content.")
    
    result = cf.run(
        "Explain entropy briefly, then write a haiku about it",
        agents=[scientist, poet]
    )
    
    return result

That’s it! The scientist explains entropy, then the poet writes a haiku. Each agent has clear instructions and they work together on a shared goal.

Building Complex Workflows

From this simple foundation, you can build sophisticated workflows. For example, the podcast workflow:

• Downloads episodes from RSS feeds
• Transcribes audio to text
• Extracts insights using AI agents
• Converts to HTML and sends emails

Each step is a composable task, and Korben tracks progress in CSV files so you can resume interrupted workflows without re-processing.

What’s Next?

Korben is under active development. Coming soon:

• More pre-built workflows
• Web dashboard for monitoring
• Enhanced configuratin and LLM support

Check out the GitHub repo to get started!

Try It Yourself

# Clone and install
git clone <repo-url>
cd korben
pdm install

# Run the entropy task
pdm run python3 ./korben.py --task entropy

The output? A scientist’s explanation of entropy followed by a poetic haiku. Simple, composable, and powerful.

We built Korben for ourselves, and we hope you find it useful too.