Deploying the Agent to Railway

This guide covers deploying the LangGraph agent to Railway.

Prerequisites

Install Railway CLI (if not already installed):
bash
```
npm install -g @railway/cli
```
Login to Railway:
bash
```
railway login
```
Initialize Railway project (from the agent directory):
bash
```
cd agent
railway init
```

Set environment variables:

bash

railway variables set GOOGLE_API_KEY=your-google-ai-api-key-here

Go to Railway Dashboard
Create New Project:
- Click "New Project"
- Select "Deploy from GitHub repo"
- Connect your GitHub account if not already connected
- Select this repository
Configure the service:
- Railway will auto-detect the Dockerfile
- Root directory: /agent
- Port: 8000 (automatically detected)
Set Environment Variables:
- Go to the "Variables" tab
- Add: GOOGLE_API_KEY = your Google AI API key
- (Optional) Add LangSmith variables for tracing
Deploy:
- Railway will automatically build and deploy
- Wait for the build to complete (usually 2-3 minutes)
Get the URL:
- Go to "Settings" > "Networking"
- Click "Generate Domain"
- Your agent will be available at: https://your-project.up.railway.app

Add this to your repository README:

markdown

[![Deploy on Railway](https://railway.app/button.svg)](https://railway.app/template/your-template-id)

After deploying the agent, update your Next.js frontend to use the Railway URL:

Update the agent URL:

typescript

const agent = new LangGraphAgent({
  agentUrl: process.env.AGENT_URL || "https://your-project.up.railway.app",
});

Add to your .env.local:

AGENT_URL=https://your-project.up.railway.app

Once deployed, verify the agent is running:

bash

curl https://your-project.up.railway.app/health

View logs: railway logs
View metrics: Railway Dashboard > Metrics tab
Add LangSmith for tracing: Set LANGCHAIN_TRACING_V2=true and LANGCHAIN_API_KEY

Railway pricing:

For production deployment, consider: