Skip to main content
Exa is a neural search engine that provides high-quality web search results optimized for AI applications. The langchain-exa integration provides tools and retrievers for searching the web and finding similar content.

​
Installation

Install the langchain-exa package: 
pip install langchain-exa

​
Setup

Get an API key from Exa and set it as an environment variable: 
export EXA_API_KEY="your-api-key"

​
Components

The integration provides three main components:
  • ExaSearchResults - Tool for neural search
  • ExaFindSimilarResults - Tool for finding similar pages
  • ExaSearchRetriever - Retriever for RAG applications

​
ExaSearchResults Tool

The ExaSearchResults tool performs neural search and returns detailed search results.

​
Basic Usage

from langchain_exa import ExaSearchResults

tool = ExaSearchResults()
results = tool.invoke({"query": "best time to visit japan", "num_results": 5})

​
Advanced Features

from langchain_exa import ExaSearchResults, TextContentsOptions

tool = ExaSearchResults()

# Search with content options
results = tool.invoke({
    "query": "artificial intelligence breakthroughs 2024",
    "num_results": 10,
    "text_contents_options": {"max_characters": 500},
    "highlights": True,
    "summary": True,
    "type": "neural",  # or "keyword" or "auto"
    "start_published_date": "2024-01-01",
    "include_domains": ["arxiv.org", "nature.com"]
})

​
Parameters

  • query - The search query (required)
  • num_results - Number of results to return (1-100, default: 10)
  • text_contents_options - Content extraction options:
    • True - Return full text
    • {"max_characters": N} - Limit text length
  • highlights - Include highlighted excerpts
  • summary - Include AI-generated summaries
  • type - Search type: "neural", "keyword", or "auto"
  • include_domains - Filter to specific domains
  • exclude_domains - Exclude specific domains
  • start_published_date - Filter by publish date (YYYY-MM-DD)
  • end_published_date - Filter by publish date (YYYY-MM-DD)
  • start_crawl_date - Filter by crawl date (YYYY-MM-DD)
  • end_crawl_date - Filter by crawl date (YYYY-MM-DD)
  • use_autoprompt - Auto-optimize query for better results
  • livecrawl - Crawl live pages: "always", "fallback", or "never"

​
ExaFindSimilarResults Tool

Find pages similar to a given URL. 
from langchain_exa import ExaFindSimilarResults

tool = ExaFindSimilarResults()
results = tool.invoke({
    "url": "https://example.com/article",
    "num_results": 5,
    "exclude_source_domain": True
})

​
ExaSearchRetriever

The retriever is designed for RAG (Retrieval-Augmented Generation) applications.

​
Basic Retriever

from langchain_exa import ExaSearchRetriever

retriever = ExaSearchRetriever(k=10)
documents = retriever.invoke("best time to visit japan")

for doc in documents:
    print(f"Title: {doc.metadata['title']}")
    print(f"URL: {doc.metadata['url']}")
    print(f"Content: {doc.page_content[:200]}...")

​
Retriever with Highlights

from langchain_exa import ExaSearchRetriever

retriever = ExaSearchRetriever(
    k=5,
    highlights=True,
    text_contents_options={"max_characters": 1000}
)

documents = retriever.invoke("machine learning research")

for doc in documents:
    highlights = doc.metadata.get("highlights", [])
    print(f"Highlights: {highlights}")

​
Retriever with Summary

from langchain_exa import ExaSearchRetriever

retriever = ExaSearchRetriever(
    k=3,
    summary=True,
    type="auto"
)

documents = retriever.invoke("climate change solutions")

for doc in documents:
    print(f"Summary: {doc.metadata['summary']}")

​
Configuration Options

The retriever supports the same parameters as ExaSearchResults: 
retriever = ExaSearchRetriever(
    k=10,  # Number of results
    include_domains=["edu", "gov"],
    exclude_domains=["example.com"],
    start_published_date="2023-01-01",
    use_autoprompt=True,
    type="neural",  # Search type
    highlights=True,
    summary=True,
    livecrawl="fallback"
)

​
Using with Agents

Integrate Exa tools with LangChain agents: 
from langchain_exa import ExaSearchResults, ExaFindSimilarResults
from langchain_openai import ChatOpenAI
from langchain.agents import create_tool_calling_agent, AgentExecutor
from langchain_core.prompts import ChatPromptTemplate

# Initialize tools
search_tool = ExaSearchResults()
similar_tool = ExaFindSimilarResults()
tools = [search_tool, similar_tool]

# Create agent
model = ChatOpenAI(model="gpt-4")
prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful research assistant."),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}")
])

agent = create_tool_calling_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools)

# Run agent
result = agent_executor.invoke({
    "input": "Search for recent AI research papers and find similar articles"
})

​
Using in RAG Chains

Combine with LangChain for retrieval-augmented generation: 
from langchain_exa import ExaSearchRetriever
from langchain_openai import ChatOpenAI
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough

# Set up retriever
retriever = ExaSearchRetriever(
    k=5,
    text_contents_options={"max_characters": 1000},
    summary=True
)

# Create RAG chain
prompt = ChatPromptTemplate.from_template(
    """Answer the question based on the following context:

{context}

Question: {question}

Answer:"""
)

model = ChatOpenAI(model="gpt-4")

def format_docs(docs):
    return "\n\n".join([doc.page_content for doc in docs])

chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

# Use the chain
answer = chain.invoke("What are the latest developments in quantum computing?")
print(answer)

​
API Reference

For detailed API documentation, see:

​
Resources

Build docs developers (and LLMs) love

Get started for freeTalk to us