Skip to main content

​
Prerequisites

  • Sign up for a Pinecone account at Pinecone.io.
  • Get your Pinecone API key from the Pinecone dashboard.

​
Setup

  1. Add your Pinecone API key to the .env file.
.env
PINECONE_API_KEY=your-pinecone-api-key
  1. The Starter-Kit already includes the necessary configuration for Pinecone in src/lib/ai/pinecone.ts, if you have selected the Pinecone Provider via the CLI. This file initializes the Pinecone client.

​
Usage

To use Pinecone in your application, you can create an API route that interacts with the Pinecone vector database. Here’s an example of how to set up a route that uses Pinecone for similarity search: 
import { kv } from '@vercel/kv';
import { pc } from '@/lib/ai/pinecone';
import { Ratelimit } from '@upstash/ratelimit';
import { NextRequest, NextResponse } from 'next/server';
// Allow responses up to 30 seconds
export const maxDuration = 30;

// Create Rate limit
const ratelimit = new Ratelimit({
    redis: kv,
    limiter: Ratelimit.fixedWindow(5, '30s'),
});

export async function POST(req: NextRequest) {
    // call ratelimit with request ip
    const { success, remaining } = await ratelimit.limit(req.ip ?? 'ip');

    // block the request if unsuccessful
    if (!success) return new Response('Rate limit exceeded', { status: 429 });

    const { vector, topK } = await req.json();

    try {
        // Assuming you have an index named 'my-index'
        const index = pc.Index(process.env.PINECONE_INDEX);

        const queryResponse = await index.query({
            vector,
            topK,
            includeMetadata: true,
        });

        return NextResponse.json({ results: queryResponse.matches });
    } catch (error) {
        console.error('Error querying Pinecone:', error);
        return NextResponse.json({ error: 'Failed to query Pinecone' }, { status: 500 });
    }
}

​
Common Operations

Pinecone offers various features for vector database management. Here are a few examples:

​
Upserting Vectors

const index = pc.Index(process.env.PINECONE_INDEX);

await index.upsert([
	{
		id: "vec1",
		values: [0.1, 0.2, 0.3, 0.4, 0.5],
		metadata: { text: "This is a sample vector" },
	},
	// Add more vectors as needed
]);

​
Deleting Vectors

const index = pc.Index(process.env.PINECONE_INDEX);

await index.deleteOne("vec1");
// Or delete multiple vectors
// await index.deleteMany(['vec1', 'vec2', 'vec3']);

​
Fetching Vectors

const index = pc.Index(process.env.PINECONE_INDEX);

const fetchResponse = await index.fetch(["vec1", "vec2"]);
console.log(fetchResponse.vectors);
See the Pinecone documentation for more information on the available operations and their usage.