This example shows how to use Workflow to make AI agents more reliable and production-ready by adding automatic retries, resume capabilities, and fault tolerance to AI SDK applications. It showcases a conversational flight booking assistant that can search flights, check status, and book tickets—all while being resilient to network failures and LLM errors.

• An API key from Vercel AI Gateway

1. Clone this example and install dependencies:

   git clone https://github.com/vercel/workflow-examples
   cd workflow-examples/flight-booking-app
   pnpm install

2. Create a .env.local file:

   touch .env.local

3. Add your API key to the .env.local file:

   AI_GATEWAY_API_KEY=your_api_key_here

4. Start the development server:

   pnpm dev

5. Open http://localhost:3000 to see the app

Deploy directly to Vercel, no additional configuration is needed. Workflow works out of the box.

For non-Vercel deployments, you can configure a PostgreSQL World to handle workflow state persistence.

1. Set up a PostgreSQL database (Supabase, Neon, etc.)

2. Add environment variables:

   WORKFLOW_TARGET_WORLD="@workflow/world-postgres"
   WORKFLOW_POSTGRES_URL="postgres://postgres:password@db.yourdb.co:5432/postgres"
   WORKFLOW_POSTGRES_JOB_PREFIX="workflow_"
   WORKFLOW_POSTGRES_WORKER_CONCURRENCY=10

3. Create the database schema:

   pnpm exec workflow-postgres-setup

4. Start the PostgreSQL World:

In your Next.js instrumentation.ts:

export async function register() {
  if (process.env.NEXT_RUNTIME !== "edge") {
    console.log("Starting workflow workers...");
    import("workflow/runtime").then(async ({ getWorld }) => {
      console.log("Starting Postgres World...");
      await getWorld().start?.();
    });
    console.log("Workflow workers started!");
  }
}

5. Deploy to your platform of choice

Learn more about the Workflow PostgreSQL World here.

• DurableAgent - @workflow/ai's DurableAgent provides automatic retries, fault tolerance, and stream reconnection for AI SDK applications
• Multi-turn conversations - Agent maintains conversation state across tool-calling loops and multiple LLM interactions
• Stream reconnection - Client can reconnect to in-progress workflows using WorkflowChatTransport after network failures
• Tool execution - Five flight booking tools (search, status check, airport info, booking, baggage) showing real-world agent patterns
• PostgreSQL World - Optional PostgreSQL backend for custom deployment needs beyond Vercel (Railway, Render, etc.)

This project uses the following stack:

• Next.js 15 (App Router)
• Vercel AI SDK with streamText and tools
• Workflow DevKit for durability
• Tailwind CSS for styling