Updated for LlamaIndex v0.10+: Building a RAG pipeline using LlamaIndex, Google Gemini Pro, and Pinecone

main.py

import os
from pinecone import Pinecone
from llama_index.llms.gemini import Gemini
from llama_index.vector_stores.pinecone import PineconeVectorStore
from llama_index.embeddings.gemini import GeminiEmbedding
from llama_index.core import StorageContext, VectorStoreIndex, download_loader

from llama_index.core import Settings

GOOGLE_API_KEY = "YOUR_API_KEY"
PINECONE_API_KEY = "YOUR_API_KEY"

os.environ["GOOGLE_API_KEY"] = GOOGLE_API_KEY
os.environ["PINECONE_API_KEY"] = PINECONE_API_KEY

DATA_URL = "https://www.gettingstarted.ai/how-to-use-gemini-pro-api-llamaindex-pinecone-index-to-build-rag-app/"

# set llm as Gemini Pro
llm = Gemini()
embed_model = GeminiEmbedding(model_name="models/embedding-001")

Settings.llm = llm
Settings.embed_model = embed_model
Settings.chunk_size = 512

# create pinecone client
pinecone_client = Pinecone(api_key=os.environ["PINECONE_API_KEY"])

# list pinecone indexes
for index in pinecone_client.list_indexes():
    print(index['name'])

# select pinecone index 'posts'
pinecone_index = pinecone_client.Index("demo")
# index_description = pinecone_client.describe_index("posts")

# load page using llamaindex
BeautifulSoupWebReader = download_loader("BeautifulSoupWebReader")

loader = BeautifulSoupWebReader()
documents = loader.load_data(urls=[DATA_URL])

# store embeddings in pinecone index
vector_store = PineconeVectorStore(pinecone_index=pinecone_index)

index = VectorStoreIndex.from_vector_store(vector_store=vector_store)

storage_context = StorageContext.from_defaults(
    vector_store=vector_store
)

# create Pinecone index programmatically 
# index = VectorStoreIndex.from_documents(
#     documents, 
#     storage_context=storage_context
# )

# query pinecone index for similar embeddings
query_engine = index.as_query_engine()
gemini_response = query_engine.query("What does the author think of LlamaIndex?")

# print response
print(gemini_response)