Skip to content

Quick Start Guide#

Installation#

To get started with MTLLM, install the base package:

pip install mtllm

MTLLM supports multiple LLM providers. Choose and install the integration you need:

pip install mtllm[openai]

pip install mtllm[anthropic]

pip install mtllm[google]

pip install mtllm[groq]

pip install mtllm[huggingface]

pip install mtllm[ollama]

pip install mtllm[together]

Your First AI Integrated Function#

Let's build a simple translation function that demonstrates how MTLLM transforms ordinary functions into intelligent, reasoning components.

The Traditional Approach#

Here's how you'd typically handle translation with manual API integration:

def translate(eng_sentence: str, target_lang: str) -> str {
    # Traditional approach: manual API calls, prompt engineering, response parsing
    # Lots of boilerplate code would go here...

    return "Hola Mundo";  # Hardcoded for demo
}

with entry {
    print(translate("Hello World", "es"));
}

The Problem: You're limited to language codes (es, fr, etc.) and need extensive prompt engineering to handle natural language inputs like "Language spoken in Somalia" or "The language of Shakespeare."

The MTP Way: by keyword#

With the by keyword abstraction in MTLLM, your functions become intelligent agents that can reason about their inputs and produce contextually appropriate outputs:

Step 1: Import Your LLM#

import from mtllm.llm {Model}

glob llm = Model(model_name="gpt-4o");

Step 2: Transform Your Function into an Agent#

Simply add by llm to make your function AI-integrated:

import from mtllm.llm {Model}

glob llm = Model(model_name="gpt-4o");

def translate(eng_sentence: str, target_lang: str) -> str by llm();

with entry {
    print(translate("Hello World", "Language spoken in Somalia"));
    print(translate("Good morning", "The language of Cervantes"));
    print(translate("Thank you", "What people speak in Tokyo"));
}

That's it! 🎉 Your function now intelligently understands natural language descriptions and performs contextual translation.

Step 3: Run Your AI Integrated Application#

Set your API key and run:

export OPENAI_API_KEY="your-api-key-here"
jac run translator.jac

Ready to explore more advanced ways of using the by abstraction? Continue with the Usage Guide to learn about all the ways you can build AI-integrated software with MTLLM, including object methods, function overriding, and complex multi-agent workflows.