Skip to content
Varity Docs

Build an App Using Only AI (No Code Required)

Varity Team Core Contributors Updated May 2026

This tutorial shows you how to build and deploy an app using only natural language prompts. You’ll use your AI editor (Cursor or Claude Code) to write the app, then use Varity’s MCP server to handle setup checks, the deploy workflow, and status checks.

What You’ll Build

Section titled “What You’ll Build”

A client invoicing dashboard with:

  • User authentication (implemented in your app, deployed as-is)
  • Database for storing invoices (auto-wired on deploy)
  • CRUD operations (create, read, update, delete)
  • Production hosting at a live URL

All without writing a single line of code yourself. Your AI editor writes the app; Varity deploys it.

Prerequisites

Section titled “Prerequisites”

Install the following (one-time setup):

  1. Cursor or Claude Code (AI editor with MCP support)

  2. Node.js 20+: download here

  3. Python 3.8+: download here

  4. Varity MCP: install with:

    Terminal window
    claude mcp add varity -- npx -y @varity-labs/mcp@beta

    Or for Cursor, add to .cursor/mcp.json:

    {
      "mcpServers": {
        "varity": {
          "command": "npx",
         "args": ["-y", "@varity-labs/mcp@beta"]
        }
      }
    }

That’s it. No complex infrastructure setup.

Step-by-Step

Section titled “Step-by-Step”

  1. Check your setup

    In Cursor or Claude Code, type:

    “Check if my Varity environment is set up correctly”

    What happens: The AI runs varity_doctor to verify Node.js, npm, and varitykit are installed.

    Expected output:

    ✓ Node.js version: 20.0.0
    ✓ npm version: 10.0.0
    ✓ varitykit CLI: installed
    ✓ All checks passed!

    If any checks fail, the AI will guide you through fixing them.

  2. Build your app with your AI editor

    Prompt:

    “Build a Next.js invoice manager called invoice-manager with a dashboard, login page, and a database for invoices”

    What happens: Your AI editor (Cursor or Claude Code) writes the app code: a Next.js project with a dashboard, login page, and the data layer for invoices. This is normal AI-assisted coding, no special Varity syntax.

    You now have a working app in your project folder, without writing any code yourself.

  3. Add your data model

    Prompt:

    “Add an invoices data model with fields: customerName (string), amount (number), dueDate (Date), status (one of: draft, sent, paid, overdue), and pages to create, list, edit, and delete invoices”

    What happens: Your AI editor adds the data model and the create, read, update, and delete pages to your project. When you deploy, Varity detects the database client in your dependencies and provisions the database for you.

  4. Install dependencies

    Prompt:

    “Install the dependencies for my project”

    What happens: The AI runs varity_install_deps to install all npm packages.

    Expected output:

    ✓ Installing dependencies...
    ✓ Installed 247 packages

  5. Preview locally

    Prompt:

    “Start the dev server and open it in my browser”

    What happens:

    • AI runs varity_dev_server with action: "start"
    • AI runs varity_open_browser with the localhost URL

    Expected output:

    ✓ Dev server running at http://localhost:3000
    ✓ Opening in default browser...

    Your app opens in the browser. You can:

    • Sign in with email (or Google/GitHub)
    • See the dashboard
    • Create, edit, and delete invoices

  6. Build for production

    Prompt:

    “Build my app for production”

    What happens: The AI runs varity_build to compile and optimize the app.

    Expected output:

    ✓ Detected framework: Next.js
    ✓ Running build...
    ✓ Build completed successfully
    ✓ Output: 2.4 MB (optimized)

  7. Deploy to production

    Prompt:

    “Deploy my app to production”

    What happens: The AI runs varity_deploy to:

    • Upload your app to distributed hosting
    • Inject production credentials (auth, database)
    • Generate a live URL

    Expected output:

    ✓ Deploying invoice-manager...
    ✓ Build: 100%
    ✓ Upload: 100%
    ✓ Live at: https://varity.app/invoice-manager-abc123/

    Your app is now live on the internet. Anyone can access it.

  8. Check deployment status

    Prompt:

    “Show me my deployment status”

    What happens: The AI runs varity_deploy_status to show all deployments.

    Expected output:

    Deployment: abc123
      URL: https://varity.app/invoice-manager-abc123/
      Status: active
      Framework: Next.js
      Size: 2.4 MB
      Created: 2 minutes ago

  9. Compare costs

    Prompt:

    “How much would this cost to host for 500 users on Varity vs Vercel?”

    What happens: The AI runs varity_cost_calculator to show cost breakdown.

    Expected output:

    Hosting costs for 500 users:
    

    Varity:
      - Profile-based monthly cost (fixed by selected deployment profile)
    Competitors:
      - Usage-based cost that grows with traffic and invocations

What Just Happened?

Section titled “What Just Happened?”

Here is what you accomplished, without writing any code yourself:

  1. ✅ Built a production app: Authentication, database, dashboard
  2. ✅ Added a custom data model: Invoices with create, read, update, and delete
  3. ✅ Deployed to production: Live URL with database and hosting configured automatically
  4. ✅ Verified deployment status: URL, framework, and runtime are visible in status output

Total time: 10-15 minutes
Total cost: One flat monthly bill, predictable on day 1 and day 1000
Code written: 0 lines

Next Steps

Section titled “Next Steps”

Now that you have a live app, you can:

  • Add more features: “Add a clients page with name, email, and company fields”
  • Customize the UI: “Change the primary color to purple”
  • View logs: “Show me deployment logs” → varity_deploy_logs
  • Scale up: “How much would 5,000 users cost?” → varity_cost_calculator

Advanced Prompts

Section titled “Advanced Prompts”

Add authentication customization:

“Allow users to sign in with Apple and Twitter in addition to email and Google”

Add a new page:

“Create a reports page that shows total invoices, paid vs unpaid, and totals this month”

Add relationships:

“Create a clients page and link each invoice to a client”

Deploy updates:

“Deploy my latest changes to production”

Behind the Scenes

Section titled “Behind the Scenes”

When you prompt your AI editor, here’s what happens:

Your PromptWhat Happens
”Build my app”Your AI editor writes the app code in your project
”Add a data model”Your AI editor adds the model and pages
”Start dev server”varity_dev_server runs npm run dev, returns localhost URL
”Build for production”varity_build compiles the app, optimizes the bundle
”Deploy”varity_deploy deploys the app and injects credentials
”Show status”varity_deploy_status fetches deployment info
”Compare costs”varity_cost_calculator compares Varity vs Vercel, Render, or Railway

Every tool runs locally or via Varity’s infrastructure. No manual configuration needed.

Troubleshooting

Section titled “Troubleshooting”

“varitykit not found”

Install the CLI:

Terminal window
pipx install varitykit
varitykit doctor

“Cannot find module @varity-labs/mcp”

Verify Node.js 20+ is installed:

Terminal window
node --version

MCP server not showing in Cursor

Restart Cursor after adding .cursor/mcp.json. Verify the JSON is valid.

Deploy fails

Ask the AI:

“Show me the build logs for my last deployment”

The AI runs varity_deploy_logs to debug.

Why This Works

Section titled “Why This Works”

Traditional app development requires:

  • Learning 5+ technologies (Next.js, React, Tailwind, Postgres, an auth provider)
  • Configuring hosting, a database add-on, and an auth service separately
  • Writing 1,000+ lines of boilerplate code
  • Managing environment variables, credentials, deployments

Varity eliminates all of that:

  • Your AI editor writes the app code with best practices baked in
  • Auto-configuration of auth, database, hosting on deploy
  • AI-powered deploy workflow via MCP tools
  • One-command deploys with automatic credential injection

Result: You can build production apps in minutes instead of weeks.

Cost Breakdown

Section titled “Cost Breakdown”

Varity charges one flat monthly cost per app, based on the hardware you reserve. Vercel, Render, and Railway bill on usage, so the same app costs more as traffic grows and you add separate database and auth add-ons.

For a small production app with a database, a typical comparison looks like:

ProviderAt launchAt traction
VarityOne flat monthly cost (fixed)Same flat cost, your bill does not change with traffic
Vercel / Render / RailwayHigher base, plus add-onsGrows with bandwidth, invocations, and usage

Overall, Varity keeps your bill flat as your app scales while usage-metered platforms ramp with traffic, and Varity handles all configuration. No DevOps engineer needed.

For exact numbers, ask your AI editor to run varity_cost_calculator, or see Pricing and Costs.

Support

Section titled “Support”

If you get stuck:

Section titled “Related”