Build Your First Multilingual AI Agent with SUTRA and Agno

SUTRA, a family of large multilingual language models (LMLMs), excels in handling over 50 languages, making it ideal for applications requiring cultural and linguistic diversity. Agno, a lightweight and model-agnostic library, complements SUTRA by enabling the creation of intelligent agents with memory, reasoning, and tool integration.

This guide will walk you through setting up a multilingual AI agent, running examples, and leveraging tools like DuckDuckGo and Yahoo Finance for real-time data.

Prerequisites

Before diving in, ensure you have:

• A SUTRA API key from TWO AI's SUTRA API page
• Basic familiarity with Python and Jupyter notebooks
• Access to Google Colab

Setting Up the Environment

Step 1: Install Required Packages

!pip install openai agno duckduckgo-search yfinance

These packages include:

• openai: For interacting with SUTRA's API.
• agno: The Agno library for building AI agents.
• duckduckgo-search: For real-time web searches.
• yfinance: For accessing stock market data.

Step 2: Configure API Keys

import os
from google.colab import userdata

# Set the API keys from Colab secrets
os.environ["OPENAI_API_KEY"] = userdata.get("SUTRA_API_KEY")
os.environ["TAVILY_API_KEY"] = userdata.get("TAVILY_API_KEY")

🔐 Replace SUTRA_API_KEY with your actual key from TWO AI.

Using SUTRA with OpenAI Client

from openai import OpenAI

client = OpenAI(
    base_url='https://api.two.ai/v2',
    api_key=os.environ["SUTRA_API_KEY"]
)

response = client.chat.completions.create(
    model="sutra-v2",
    messages=[
        {"role": "system", "content": "You are a helpful assistant that specializes in Indian languages and culture."},
        {"role": "user", "content": "Tell me about the importance of the Ganga river in Indian culture."}
    ]
)

print(response.choices[0].message.content)

Building Multilingual AI Agents with Agno

Example 1: Cultural Q&A Agent

from agno.agent import Agent
from agno.models.openai.like import OpenAILike

sutra_agent = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    description="You are a helpful assistant specializing in Indian culture and history.",
    markdown=True
)

sutra_agent.print_response("Tell me about the history of yoga in India.", stream=True)

Example 3: Storytelling Agent

story_agent = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    description="You are a creative storyteller specializing in Indian folklore.",
    markdown=True
)

story_agent.print_response("Write a short story about a magical tree in an Indian village.", stream=True)

Example 2: Code Explanation Agent

sample_code = '''
def diwali_date(year):
    return f'Diwali in {year} is likely in October or November.'
print(diwali_date(2025))
'''

code_agent = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    description="You are an expert in explaining Python code, especially related to Indian culture.",
    markdown=True
)

code_agent.print_response(f"Explain this Python code:\n{sample_code}", stream=True)

Example 3: Stock Market Agent

from agno.tools.yfinance import YFinanceTools

sutra_agent = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    description="You are an expert in analyzing stock market data using YFinanceTools.",
    markdown=True,
    tools=[YFinanceTools()]
)

sutra_agent.print_response("How is TSLA stock doing right now in Hindi?", stream=True)

Example 4: Multilingual Agent with Reasoning

from agno.tools.duckduckgo import DuckDuckGoTools

sutra_agent_with_tools = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    description="You are a helpful assistant specializing in Indian languages, culture, and current events. Provide accurate and detailed responses in Hindi when requested, using DuckDuckGoTools for up-to-date information.",
    tools=[DuckDuckGoTools()],
    show_tool_calls=True,
    markdown=True
)

query = '''
भारत के अंतरिक्ष कार्यक्रम में हाल के विकास क्या हैं? DuckDuckGoTools का उपयोग करके नवीनतम जानकारी प्राप्त करें और निम्नलिखित शामिल करें:
1. हाल की मिशन सफलताएँ (उदाहरण: SpaDeX, Gaganyaan)।
2. भविष्य की योजनाएँ (उदाहरण: चंद्रयान-4, भारतीय अंतरिक्ष स्टेशन)।
3. निजी क्षेत्र की भागीदारी।
हिंदी में विस्तृत और संरचित उत्तर प्रदान करें, प्रत्येक अनुभाग को स्पष्ट रूप से लेबल करें।
'''

sutra_agent_with_tools.print_response(query, stream=True)

Next Steps

• 🔍 Experiment with Prompts: Customize prompts and agent descriptions.
• 🧠 Combine Tasks: Build agents that merge tasks like translation + storytelling.
• 📘 Explore the API: Visit the Sutra API Docs for more options.
• 🌍 Share Your Work: Publish your creations and tag the community!

🌟 Share Your Work

Contribute your chatbot to the open-source community:

• ✨ Submit to sutra-cookbook GitHub repo
• Share your notebook with your team or Audience

💡 Tips & Tricks

• ✅ Multilingual Power: Use sutra-v2 to support 50+ languages
• 📚 Optimal Chunks: Stick with chunk_size = 1000 and chunk_overlap = 100
• 🌍 Community First: Star the repo and share feedback

📘️ Conclusion

Combining RAG with SUTRA empowers you to build intelligent, multilingual, document-aware chatbots—perfect for education, community discussions, and global learning.

🔗 Resources & Community

• 🌐 Website: two.ai
• 💻 GitHub: sutra-cookbook
• 💬 Discord: Join the community
• 🤞 Twitter: @sutra_dev
• 💼 LinkedIn: TWO Platforms