Pinecone | Documentation

The Pinecone Data Connector allows users to store and query text embeddings as vectors. Users can then perform similarity search operations on stored vectors.

#Release stage

Alpha

#Configuration

There are two things needed to configure a Pinecone connector: Pinecone URL (corresponding to your index) and an API key. Firstly, users need to create an index on https://www.pinecone.io. Once an index is created, the URL and API key can be used to configure a connector as mentioned below - For the up-to-date configuration, refer to here.

Field	Type	Note
url*	string	Fill your Pinecone url here
api_key*	string	Fill your Pinecone api_key here

#No-code Setup

To create a Pinecone data connector effortlessly, follow these steps:

Go to the Resource page and click Add Resource
Select Pinecone
Fill in the required fields
Optionally, give a short description in the Description field

#Low-code Setup

For advanced users who prefer a more hands-on approach, you can create a Pinecone data connector using the following cURL commands:

Create a Pinecone data connector:

cURL

curl  --location http://localhost:8080/vdp/v1alpha/users/<user-id>/connector-resources
      --header 'Authorization: Bearer <Your-API-Token>' \
      --data '{
        "id": "pinecone",
        "connector_definition_name": "connector-definitions/data-pinecone",
        "description": "The Pinecone vector database",
        "configuration": {
          "url": <Pinecone host address>,
          "api_key": <api key>
        }
      }'

For other operations, please refer to the VDP Protobufs.

#Supported Tasks

The Pinecone connector supports two tasks: Upsert and Query.

Upsert - use this to insert a new vector embedding or modify existing ones. It accepts an array of vectors values as a required parameter. Where each vector is an object with an id and an array of floating point numbers.

Ref: https://docs.pinecone.io/reference/upsert

{
    "id": 1,
    "values": [1.35, -1.3, 1.356, 0.356, ...]
}

Query - use this to perform a similarity search on stored vectors. It requires 2 parameters: top_k - the number of matches to return, and vector - the query vector which is an array of floating point numbers.

Ref: https://docs.pinecone.io/reference/query

{
  "top_k": 5,
  "vector": [1.35, -1.3, 1.356, 0.356, ...]
}