• ✅ Create an index

• ✅ Convert a document into embeddings

• ✅ Upsert vectors with IDs

• ✅ Query similar content

🧪 Prerequisites

Make sure you have the following installed:

pip install pinecone-client langchain openai

Also, set your environment variables (via .env or manually):

export PINECONE_API_KEY=your_pinecone_key
export OPENAI_API_KEY=your_openai_key

📘 1. Create the Pinecone Index (once)

import os
import pinecone
from pinecone import ServerlessSpec

# Load your API key
pinecone.init(api_key=os.getenv("PINECONE_API_KEY"))

# Define index name and embedding dimension
index_name = "demo-index"
dimension = 1536  # For OpenAI embeddings

# Check if it exists
if index_name not in [i.name for i in pinecone.list_indexes()]:
    pinecone.create_index(
        name=index_name,
        dimension=dimension,
        metric="cosine",
        spec=ServerlessSpec(cloud="aws", region="us-east-1"),
    )

# Connect to the index
index = pinecone.Index(index_name)

🧠 2. Generate Embeddings using OpenAI

from langchain.embeddings import OpenAIEmbeddings

embedding_model = OpenAIEmbeddings()

texts = [
    "AI is transforming the world.",
    "Pinecone helps you do similarity search.",
    "LangChain connects LLMs with data and tools.",
]

# Embed texts
embeddings = embedding_model.embed_documents(texts)

📥 3. Upsert Vectors with IDs

# Create ID-value pairs
to_upsert = [
    {"id": f"text-{i}", "values": vector, "metadata": {"content": texts[i]}}
    for i, vector in enumerate(embeddings)
]

# Upload to Pinecone
index.upsert(vectors=to_upsert)

🔍 4. Query Similar Text

query_text = "What does Pinecone do?"
query_embedding = embedding_model.embed_query(query_text)

# Search
result = index.query(vector=query_embedding, top_k=2, include_metadata=True)

# Display result
for match in result.matches:
    print(f"Score: {match.score:.2f}")
    print("Matched content:", match.metadata['content'])

🧼 5. Clean up (optional)

# Delete the index if needed
# pinecone.delete_index(index_name)

📊 Output Example

Score: 0.87
Matched content: Pinecone helps you do similarity search.

Score: 0.63
Matched content: LangChain connects LLMs with data and tools.