Deploying to Vercel Edge

The "Edge" means running your code on thousands of servers globally, close to the user. For AI, this reduces latency and (often) increases timeout limits for streaming.

Configuration

In Next.js App Router, simply add this export:

// app/api/chat/route.ts
export const runtime = 'edge'; // <--- The Magic Line

export async function POST(req: Request) {
  // ...
}

Limitations of Edge

Edge environments are NOT Node.js.

• ❌ No fs (FileSystem) access.
• ❌ No native modules (like some sharp image resizing tools).
• ❌ Limited database connections (Need connection pooling or HTTP-based drivers like Neon/Supabase).

Streaming Response

Vercel Edge is optimized for Response objects with streams.

import { OpenAIStream, StreamingTextResponse } from 'ai';

// ... call openai ...
const stream = OpenAIStream(response);
return new StreamingTextResponse(stream);

Environment Variables

1. Go to Vercel Dashboard -> Settings -> Environment Variables.
2. Add OPENAI_API_KEY.
3. Ensure it is available in Production and Preview.

Troubleshooting

Error: "The edge function crashed"

• Did you try to use fs.readFileSync?
• Did you import a huge library? (Edge functions have size limits, usually < 1MB).

Solution: If you need heavy libraries, switch back to runtime = 'nodejs'.