SUTRA with AGNO

SUTRA by TWO Platforms

SUTRA is a family of large multi-lingual language models (LMLMs) pioneered by Two Platforms. SUTRA’s dual-transformer approach extends the power of both MoE and Dense AI language model architectures, delivering cost-efficient multilingual capabilities for over 50+ languages. It powers scalable AI applications for conversation, search, and advanced reasoning, ensuring high-performance across diverse languages, domains and applications.

What is Agno?

Agno is a lightweight library for building agents with memory, knowledge, tools, and reasoning capabilities. It's model-agnostic, allowing you to connect to 23+ model providers without lock-in.

Get Your API Keys

Before you begin, make sure you have:

A SUTRA API key (Get yours at TWO AI's SUTRA API page)
Basic familiarity with Python and Jupyter notebooks

This notebook is designed to run in Google Colab, so no local Python installation is required.

Setup and Installation

First, let's install the required packages:

# SUTRA models are OpenAI API compatible
!pip install openai agno duckduckgo-search

Setting up Environment Variables

You'll need to set up your API keys. For security reasons, it's best to use environment variables:

import os
from google.colab import userdata

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

Basic Usage of SUTRA with OpenAI Client

Let's first see how to use SUTRA with the standard OpenAI client:

from openai import OpenAI

# Initialize the client with Sutra's API endpoint
client = OpenAI(
    base_url='https://api.two.ai/v2',
    api_key=os.environ["SUTRA_API_KEY"]
)

# Simple completion with Sutra
response = client.chat.completions.create(
    model="sutra-v2",  # Use Sutra model
    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)

Multilingual Capabilities of SUTRA

One of Sutra's strengths is its multilingual capabilities. Let's test it with Hindi:

# Hindi example
response = client.chat.completions.create(
    model="sutra-v2",
    messages=[
        {"role": "system", "content": "आप एक सहायक AI हैं जो हिंदी में उत्तर देती है।"},
        {"role": "user", "content": "भारतीय संस्कृति में गंगा नदी का क्या महत्व है?"}
    ]
)

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

Integrating SUTRA with Agno Agent

Now, let's integrate Sutra with Agno to create an intelligent agent. We'll create a custom OpenAI model class for Agno that uses Sutra:

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

# Initialize the Agent with Sutra model via OpenAILike wrapper
agent = Agent(
    model=OpenAILike(
        id="sutra-v2",
        api_key=os.getenv("SUTRA_API_KEY"),
        base_url="https://api.two.ai/v2"
    ),
    markdown=True
)

# Create a basic Sutra agent
sutra_agent = Agent(
    description="You are a helpful assistant that specializes in Indian languages and culture.",
    markdown=True
)

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

Adding Tools to the SUTRA Agent

Let's enhance our Sutra agent by adding tools, such as web search capabilities:

from agno.tools.duckduckgo import DuckDuckGoTools

# Create a Sutra agent with web search capability
sutra_agent_with_tools = Agent(
    description="You are a helpful assistant that specializes in Indian languages, culture, and current events.",
    tools=[DuckDuckGoTools()],
    show_tool_calls=True,
    markdown=True
)

# Test the agent with a query that might require web search
sutra_agent_with_tools.print_response("What are the recent developments in India's space program in hindi?", stream=True)

Creating a Multilingual SUTRA Agent with Reasoning

Now, let's create a more advanced SUTRA agent that can handle multiple languages and has reasoning capabilities:

from agno.tools.reasoning import ReasoningTools

# Create a multilingual Sutra agent with reasoning capabilities
multilingual_sutra_agent = Agent(
    description="You are a multilingual assistant that can communicate in various Indian languages and has strong reasoning abilities.",
    tools=[
        ReasoningTools(add_instructions=True),
        DuckDuckGoTools()
    ],
    instructions=[
        "You can respond in the language used by the user.",
        "Use reasoning to analyze complex questions before answering.",
        "If needed, search the web for current information."
    ],
    show_tool_calls=True,
    markdown=True
)

# Test with a complex reasoning task in English
multilingual_sutra_agent.print_response(
    "Analyze the impact of climate change on agricultural practices in India. What adaptations are farmers making?",
    stream=True,
    show_full_reasoning=True
)

# Test with a Hindi query
multilingual_sutra_agent.print_response(
    "भारत में जलवायु परिवर्तन का कृषि पर क्या प्रभाव पड़ रहा है? किसान कैसे अनुकूलन कर रहे हैं?",
    stream=True,
    show_full_reasoning=True
)

Conclusion

In this notebook, we've demonstrated how to use the Sutra model with the OpenAI client and integrate it with Agno to create intelligent agents. We've explored:

Basic usage of Sutra with the OpenAI client
Multilingual capabilities of SUTRA
Creating a simple Sutra agent with Agno
Adding tools like web search to the SUTRA agent
Building a multilingual SUTRA agent with reasoning capabilities

Sutra's strong multilingual capabilities, especially for Indian languages, combined with Agno's flexible agent framework, provide a powerful platform for building intelligent applications that can understand and respond in multiple languages and leverage various tools and knowledge sources.

SUTRA with AGNO

On this page