Pinecone is a managed cloud platform for creating vector databases, called indexes, and supporting similarity-search workflows. Its Canopy framework connects this cloud setup to retrieval-augmented generation (RAG), reducing setup and coding while trading away some control. section 11.3.2
A Canopy-backed Pinecone setup requires:
- Python 3.11 to install Canopy, followed by the canopy-sdk package. section 11.3.2
- OpenAI and Pinecone API keys, plus a Pinecone account upgraded to a standard plan so Canopy can create the required index. section 11.3.2
- A data directory that Canopy can upload into an index. section 11.3.2
Testing a custom RAG index
Upload custom data with Canopy, then query it to assess retrieval in a target context. A booking dataset can produce responses covering room availability, booking creation, booking retrieval, and booking summaries, with the response identifying the source data directory. section 11.3.3
export PINECONE_API_KEY="<PINECONE_API_KEY>" export OPENAI_API_KEY="<OPENAI_API_KEY>" export INDEX_NAME="<INDEX_NAME>"
Set the Pinecone API key, OpenAI API key, and index name as environment variables before using the framework. section 11.3.2


