Pinecone cloud vector store in LangChain - Mini Project: Build & Apply

Using Pinecone Cloud Vector Store with Langchain

📖 Scenario: You want to build a simple app that stores and searches text data using Pinecone cloud vector store with Langchain. This helps you find similar texts quickly.

🎯 Goal: Create a Python script that connects to Pinecone, sets up an index, adds text vectors, and queries for similar texts.

📋 What You'll Learn

Create a Pinecone index with name example-index

Set a variable api_key with your Pinecone API key string

Use Langchain's Pinecone vector store to add vectors from texts

Query the index for similar texts using a query vector

💡 Why This Matters

🌍 Real World

Many apps use vector stores like Pinecone to quickly find similar documents, images, or texts. This project shows how to connect and use Pinecone with Langchain for text similarity search.

💼 Career

Knowing how to use cloud vector stores is useful for roles in AI, data science, and software development where fast similarity search is needed.

1

Set up Pinecone API key and initialize client

Create a variable called api_key and set it to the string 'your-pinecone-api-key'. Then import pinecone and initialize the Pinecone client with pinecone.init(api_key=api_key, environment='us-west1-gcp').

LangChain

# Create api_key variable and initialize Pinecone client
# Your code here

Need a hint?

Use pinecone.init() with your api_key and environment string.

2

Create Pinecone index named 'example-index'

Create a Pinecone index named 'example-index' with dimension 1536 using pinecone.create_index('example-index', dimension=1536). Then connect to this index with index = pinecone.Index('example-index').

LangChain

api_key = 'your-pinecone-api-key'
import pinecone
pinecone.init(api_key=api_key, environment='us-west1-gcp')
# Create Pinecone index and connect to it
# Your code here

Need a hint?

Use pinecone.create_index() before connecting with pinecone.Index().

3

Add text vectors to Pinecone using Langchain Pinecone vector store

Import Pinecone from langchain.vectorstores and OpenAIEmbeddings from langchain.embeddings. Create an embeddings object with OpenAIEmbeddings(). Then create a Pinecone vector store with vector_store = Pinecone(index, embeddings.embed_query, 'example-index'). Finally, add texts ['Hello world', 'Langchain is cool'] with IDs ['1', '2'] using vector_store.add_texts(texts, ids=ids).

LangChain

api_key = 'your-pinecone-api-key'
import pinecone
pinecone.init(api_key=api_key, environment='us-west1-gcp')
pinecone.create_index('example-index', dimension=1536)
index = pinecone.Index('example-index')
# Import Langchain classes and add texts to Pinecone
# Your code here

Need a hint?

Use add_texts method of the Pinecone vector store to add your texts with IDs.

4

Query Pinecone index for similar texts

Create a query vector by embedding the text 'Hello' using query_vector = embeddings.embed_query('Hello'). Then query the vector store with results = vector_store.similarity_search_by_vector(query_vector, k=2) to get top 2 similar texts.

LangChain

api_key = 'your-pinecone-api-key'
import pinecone
pinecone.init(api_key=api_key, environment='us-west1-gcp')
pinecone.create_index('example-index', dimension=1536)
index = pinecone.Index('example-index')
from langchain.vectorstores import Pinecone
from langchain.embeddings import OpenAIEmbeddings
embeddings = OpenAIEmbeddings()
vector_store = Pinecone(index, embeddings.embed_query, 'example-index')
texts = ['Hello world', 'Langchain is cool']
ids = ['1', '2']
vector_store.add_texts(texts, ids=ids)
# Create query vector and search for similar texts
# Your code here

Need a hint?

Use embed_query to create the vector and similarity_search_by_vector to find matches.