​
Introduction

Integrate Tavily with Composio to enhance your AI workflows with powerful web search capabilities. Composio provides a platform to connect your AI agents to external tools like Tavily, making it easy to incorporate real-time web search and data extraction into your applications.

​
Step-by-Step Integration Guide

​
Step 1: Install Required Packages

Install the necessary Python packages:

pip install composio-openai openai

​
Step 2: Set Up API Keys

Set these as environment variables in your terminal or add them to your environment configuration file:

export OPENAI_API_KEY=your_openai_api_key
export COMPOSIO_API_KEY=your_composio_api_key

​
Step 3: Connect Tavily to Composio

Add Tavily as an integration in Composio:

composio add tavily

When prompted, enter your Tavily API key:

> Enter API Key: your_tavily_api_key

​
Step 4: Example Use Case

from composio_openai import ComposioToolSet, Action
from openai import OpenAI

# Initialize OpenAI client with API key
client = OpenAI()

# Initialize Composio toolset
toolset = ComposioToolSet()

# Get the Tavily tool with all available parameters
tools = toolset.get_tools(actions=[Action.TAVILY_TAVILY_SEARCH])

# Define the market research task with specific parameters
task = {
    "query": "Analyze the competitive landscape of AI-powered customer service solutions in 2024",
    "search_depth": "advanced",  
    "include_answer": True,      
    "max_results": 10,  
    # Focus on relevant industry sources         
    "include_domains": [        
        "techcrunch.com",
        "venturebeat.com",
        "forbes.com",
        "gartner.com",
        "marketsandmarkets.com"
    ],
}

# Send request to LLM
messages = [{"role": "user", "content": str(task)}]

response = client.chat.completions.create(
    model="gpt-4",
    messages=messages,
    tools=tools,
    tool_choice="auto"
)

# Handle tool call via Composio
execution_result = None
response_message = response.choices[0].message

if response_message.tool_calls:
    execution_result = toolset.handle_tool_calls(response)
    print("Execution Result:", execution_result)
    messages.append(response_message)
    
    # Add tool response messages
    for tool_call, result in zip(response_message.tool_calls, execution_result):
        messages.append({
            "role": "tool",
            "content": str(result),
            "tool_call_id": tool_call.id
        })
    
    # Get final response from LLM
    final_response = client.chat.completions.create(
        model="gpt-4",
        messages=messages
    )
    
    print("\nMarket Research Summary:")
    print(final_response.choices[0].message.content)
else:
    print("LLM responded directly (no tool used):", response_message.content)

​
Additional Use Cases

  1. Research Automation: Automate the collection and summarization of research data
  2. Content Curation: Gather and organize information from multiple sources
  3. Real-time Data Integration: Keeping your AI models up-to-date with the latest information.