Documentation

Learn how to integrate vector search and RAG capabilities into your Next.js application.

Quick Start

Follow the setup instructions below to get started with vector search in minutes. Make sure you have PostgreSQL with pgvector extension enabled.

Getting Started

Complete these steps to set up vector search in your application.

Step 1

Database Setup

Create a Postgres database with either:

Vercel Postgres (recommended for quick start)
NeonDB (recommended for production)

Step 2

Environment Variables

1# Database
2POSTGRES_URL="postgres://..."
3
4# OpenAI
5OPENAI_API_KEY="sk-..."
6
7# Inngest (optional, for background jobs)
8INNGEST_EVENT_KEY="..."
9INNGEST_SIGNING_KEY="..."

Step 3

Enable pgvector

Connect to your database and run:

1CREATE EXTENSION vector;

Step 4

Create Tables

Using Prisma schema:

1model items {
2  id        BigInt                 @id @default(autoincrement())
3  embedding Unsupported("vector")?
4  metadata  Json?                  @default("{}")
5  createdAt DateTime              @default(now())
6  updatedAt DateTime              @updatedAt
7}

Need help? Check out our GitHub repository or join our community.