Using Tavily Search Tool

In this guide, you’ll learn how to integrate Tavily—a search API built for large language models (LLMs) and retrieval-augmented generation (RAG)—into your workflows. Tavily provides real-time, up-to-date context by connecting your LLM directly to the web.

What Is Tavily?

Tavily.com offers a fast, persistent search engine that you can call via API. Think of it as search-augmented retrieval, supplying fresh information your LLM can consume to improve accuracy and relevance.

Sign Up and Obtain Your API Key

1. Go to Tavily.com and create your account.
2. Navigate to the dashboard to view your plan and API usage.

Setting Your API Key

Export your key as an environment variable before running any code.

Install the LangChain Community Tool

Make sure you have the required package:

pip install langchain-community

Using TavilySearchResults in Python

Below is a sample script that queries Tavily and prints out the results:

import os
from langchain_community.tools.tavily_search import TavilySearchResults

# Load your API key from the environment
TAVILY_API_KEY = os.getenv("TAVILY_API_KEY")

# Initialize the Tavily search tool
tool = TavilySearchResults()

# Perform a search query
response = tool.invoke({"query": "When is ICC Men's T20 World Cup 2024 starting?"})

# Inspect the results
print("Number of results:", len(response))
print("First URL:", response[0]["url"])
print("First snippet:", response[0]["content"])

Response Structure

Tavily returns a list of dictionaries with the following fields:

Example output:

Number of results: 5
First URL: https://www.espncricinfo.com/series/icc-men-s-t20-world-cup-2024-1411166/match-schedule-fixtures-and-results
First snippet: "Get 2024 T20 World Cup schedule, fixtures, scorecard updates, and results on ESPNcricinfo. Track latest match scores, schedule, and results of ICC Men's T20 World Cup 2024."

Next Steps

• Combine multiple snippets into a single context string for your LLM.
• Explore additional LangChain Community tools for advanced workflows.
• Experiment with different queries to see how real-time search data improves your applications.

Links and References

• Tavily Documentation
• LangChain Community
• Retrieval-Augmented Generation (RAG)