Documentation
ReferenceAuthoring Portal ReferenceIntegrationsRAG/Content Retrieval IntegrationsPinecone

Defining a Pinecone RAG

Important

This simplified Pinecone RAG definition will only work for Pinecone setups where the source text chunk that the vector is based on is contained within the retrieved Pinecone metadata.

A Pinecone RAG Integration Definition contains only boilerplate fields. Specific fields controlling the details of RAG retrieval are specified where the definition is used in your Agent, enabling specialization by context.

FieldDetails
NameA descriptive name to help you distinguish this integration
Pinecone Index NameThe name of the index in Pinecone. Note that this is NOT the Pinecone Host.
Access Key SecretThe iostack Secret containing the webhook URL provided by Pinecone. This must be preconfigured in the Account Secrets & Access Keys
NamespaceOptional - required if you need to specify a specific namespace within your Pinecone index.
Text Field NameThe name of the metadata field containing the text chunk that the vector is calculated from.

If you would like to create your own Pinecone RAG Index, iostack has a Google Colab Python workbook available that can be used to relatively easily configure and build your own Pinecone index.