How to integrate ai sdk for embeddings on Vercel

Navigate to your project directory and install the required packages:

npm install ai @ai-sdk/openai

This installs the core AI SDK and OpenAI provider for generating embeddings.

Create a .env.local file in your project root and add your API key:

OPENAI_API_KEY=your_openai_api_key_here

Make sure to add .env.local to your .gitignore file to keep your API key secure.

Create a new file app/api/embeddings/route.ts (for App Router) or pages/api/embeddings.ts (for Pages Router):

import { embed } from 'ai';
import { openai } from '@ai-sdk/openai';

export async function POST(req: Request) {
  const { text } = await req.json();
  
  const { embedding } = await embed({
    model: openai.embedding('text-embedding-3-small'),
    value: text,
  });
  
  return Response.json({ embedding });
}

This creates an endpoint that accepts text and returns embeddings.

Create a component or page to use the embeddings API:

const generateEmbedding = async (text: string) => {
  const response = await fetch('/api/embeddings', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({ text }),
  });
  
  const { embedding } = await response.json();
  return embedding;
};

This function sends text to your API route and receives the embedding vector.

In your Vercel dashboard, go to your project settings and click Environment Variables. Add your API key:

• Key: OPENAI_API_KEY
• Value: Your actual API key
• Environment: Production, Preview, and Development

Click Save to store the variable securely.

Start your development server to test the integration:

npm run dev

Test your embedding endpoint using a tool like curl or your frontend:

curl -X POST http://localhost:3000/api/embeddings \
  -H "Content-Type: application/json" \
  -d '{"text":"Hello, world!"}'

Verify that you receive a numerical array representing the embedding.

Deploy your project using the Vercel CLI or by pushing to your connected Git repository:

vercel --prod

Or commit and push your changes if using Git integration. Vercel will automatically build and deploy your project with the AI SDK embeddings functionality.

For production use, consider storing embeddings in a vector database. Install a vector database client:

npm install @pinecone-database/pinecone

Modify your API route to store embeddings:

// After generating embedding
await vectorDb.upsert({
  id: generateId(),
  values: embedding,
  metadata: { text }
});

This enables similarity search and retrieval capabilities.