LaunchAI Pro

Next.js 15 AI SaaS starter: Google login (Auth.js), PostgreSQL + Prisma, an OpenAI-powered AI Writer, and Stripe subscriptions.

Requirements

• Node.js 18.18 or newer (20 LTS recommended) and npm 9+
• A PostgreSQL database (Neon, Supabase or local)
• Optional, per feature: Google OAuth client, OpenAI API key, Stripe account (see "Feature → required keys")

Quick start (local)

npm ci
cp .env.example .env      # Windows: copy .env.example .env
# edit .env: set DATABASE_URL, AUTH_SECRET, NEXTAUTH_URL (see below)
npx prisma migrate deploy
npm run dev               # http://localhost:3000

npm ci installs the exact versions in package-lock.json and does not need --legacy-peer-deps. It also runs prisma generate automatically.

Never commit .env.

Environment variables

Variable	Required	What it is	Where to get it
DATABASE_URL	Yes	PostgreSQL connection string	Neon/Supabase dashboard; add ?sslmode=require for hosted databases
AUTH_SECRET	Yes	Signs session tokens	openssl rand -base64 32
NEXTAUTH_URL	Yes	Public URL of the app	http://localhost:3000 locally, your domain on Vercel
AUTH_GOOGLE_ID	For Google login	Google OAuth client ID	Google Cloud Console → APIs & Services → Credentials
AUTH_GOOGLE_SECRET	For Google login	Google OAuth client secret	Same place
OPENAI_API_KEY	For AI Writer	API key for OpenAI or any OpenAI-compatible provider	platform.openai.com → API keys, or your provider's dashboard (e.g. openrouter.ai/keys)
OPENAI_BASE_URL	No	API base URL. Leave empty for OpenAI	Your provider's docs, e.g. https://openrouter.ai/api/v1
AI_MODEL	No	Model name sent to the provider. Default gpt-4o-mini	Your provider's model list, e.g. openai/gpt-4o-mini on OpenRouter
AI_FALLBACK_MODELS	No	Comma-separated models tried in order when AI_MODEL returns 429 or 404	Same model list
STRIPE_SECRET_KEY	For billing	Stripe secret key (sk_...)	Stripe Dashboard → Developers → API keys
STRIPE_WEBHOOK_SECRET	For billing	Webhook signing secret (whsec_...)	Stripe Dashboard → Webhooks (or stripe listen locally)
STRIPE_PRICE_STARTER	For billing	Price ID of the Starter plan (price_...)	Stripe Dashboard → Product catalog
STRIPE_PRICE_PRO	For billing	Price ID of the Pro plan (price_...)	Same place
STRIPE_PRICE_ENTERPRISE	No	Price ID of the Enterprise plan. If empty, Enterprise is not sold online	Same place
SALES_CONTACT_URL	No	Enterprise "Contact sales" link, https://... or mailto:.... If empty, the card shows "Available on request"	Your contact page or sales email

All Stripe variables are read on the server at request time. No NEXT_PUBLIC_* variable and no publishable key are needed (checkout redirects to the Stripe-hosted page). After changing them on Vercel, redeploy.

Accepted aliases, useful when a host already defines them: GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET (for AUTH_GOOGLE_ID / AUTH_GOOGLE_SECRET), NEXTAUTH_SECRET (for AUTH_SECRET). Set either name, not both.

Feature → required keys

Feature	Needs
Pages /, /pricing, /login, /register	DATABASE_URL, AUTH_SECRET, NEXTAUTH_URL
Dev login (local only)	the three above, npm run dev
Google login	AUTH_GOOGLE_ID, AUTH_GOOGLE_SECRET
AI Writer (/writer, history, /api/ai/generate)	login, DATABASE_URL, OPENAI_API_KEY
Billing (/billing, checkout)	STRIPE_SECRET_KEY, STRIPE_PRICE_STARTER, STRIPE_PRICE_PRO (optional STRIPE_PRICE_ENTERPRISE, SALES_CONTACT_URL)
Subscription sync (webhook)	STRIPE_SECRET_KEY, STRIPE_WEBHOOK_SECRET

Without OPENAI_API_KEY the /writer page shows the notice "AI generation needs an API key. See README." and disables the Generate button; the API returns HTTP 503. Without STRIPE_SECRET_KEY, /billing shows the badge "Demo UI — Connect Stripe Keys to Activate"; checkout in development redirects to the dashboard with a mock success, and in production it shows "Billing is not configured".

Using another AI provider (OpenRouter and others)

The AI Writer uses the OpenAI SDK, so it works with any OpenAI-compatible provider. Set the key, the base URL and a model name from that provider. Example for OpenRouter:

OPENAI_API_KEY="sk-or-..."
OPENAI_BASE_URL="https://openrouter.ai/api/v1"
AI_MODEL="openai/gpt-4o-mini"

Leave OPENAI_BASE_URL and AI_MODEL empty to use OpenAI with gpt-4o-mini. Restart the server after changing them.

OpenRouter model slugs change over time, especially the free (:free) ones. If /writer shows "Model unavailable", pick a current slug from https://openrouter.ai/models and update AI_MODEL.

To keep /writer working when a model is rate limited (429) or removed (404), list backup models in AI_FALLBACK_MODELS, separated by commas. Each one is tried once, in order; other errors are not retried:

AI_FALLBACK_MODELS="nvidia/nemotron-3-super-120b-a12b:free,inclusionai/ling-3.1-flash"

The example uses free models only. Paid models via OpenRouter (for example openai/gpt-4o-mini) need credits on your OpenRouter account; without credits they fail with "No credits".

Provider errors are shown in /writer as clear messages: 404 → "Model unavailable", 402 → "No credits", 429 → "Rate limited", 401/403 → API key rejected.

Database setup (Neon / PostgreSQL)

1. Create a project at neon.tech (or any PostgreSQL provider) and copy the connection string.
2. Put it in .env as DATABASE_URL.
3. Create the tables:

npx prisma migrate deploy

This applies the migration in prisma/migrations/. Use npm run db:studio to browse the data.

Existing database created earlier with prisma db push? Mark the initial migration as applied once, instead of running migrate deploy:

npx prisma migrate resolve --applied 20261010000000_init

Neon users: run prisma migrate commands with Neon's direct connection string (the host without -pooler). Pooled connections can hold Prisma's migration lock or keep stale session settings and cause "advisory lock" timeouts. The pooled string is fine for the running app, so you can use the pooled URL in Vercel and the direct one only on your computer when migrating.

If you change prisma/schema.prisma, create a new migration with npx prisma migrate dev --name your_change against a development database. Do not run migrate dev against production.

Dev login (development only)

When started with npm run dev, the login page offers a "Dev Login" form so you can try the app without Google keys:

• Email: [email protected]
• Password: dev1234

On first login this creates a matching user row in the database. The provider is only registered when NODE_ENV is development, so it is not available in npm run build && npm start or on Vercel. Change or remove it in lib/auth.ts if you ship a fork.

Google OAuth

1. Google Cloud Console → APIs & Services → Credentials → Create credentials → OAuth client ID → Web application.
2. Authorized redirect URIs, add both:
   • http://localhost:3000/api/auth/callback/google
   • https://YOUR-DOMAIN/api/auth/callback/google (for example https://your-app.vercel.app/api/auth/callback/google)
3. Copy the client ID and secret into AUTH_GOOGLE_ID and AUTH_GOOGLE_SECRET.
4. NEXTAUTH_URL must match the domain you are opening. invalid_client means the ID or secret is missing or wrong; redirect_uri_mismatch means the URI above is not registered.

Stripe

1. Create recurring prices in Stripe (Starter and Pro, optionally Enterprise) and put their IDs in STRIPE_PRICE_STARTER, STRIPE_PRICE_PRO and STRIPE_PRICE_ENTERPRISE.
2. Put your secret key in STRIPE_SECRET_KEY. Use test keys (sk_test_...) until you are ready to go live.
3. Webhook endpoint: https://YOUR-DOMAIN/api/webhooks/stripe. Subscribe to exactly these events (the handler ignores all others):
   • checkout.session.completed — marks the subscription active
   • customer.subscription.updated — syncs status, price and renewal date
   • customer.subscription.deleted — downgrades the user (status CANCELED)
4. Copy the endpoint's signing secret into STRIPE_WEBHOOK_SECRET. The "Manage billing" button on /billing opens the Stripe Customer Portal. Activate it once in Stripe Dashboard → Settings → Billing → Customer portal (separately for test and live mode). Subscribed users change or cancel their plan there; the app never starts a second checkout for a user who already has a subscription.
5. Local testing with the Stripe CLI:

stripe listen --forward-to localhost:3000/api/webhooks/stripe

The CLI prints a whsec_... secret; use it as STRIPE_WEBHOOK_SECRET locally.

Deploy to Vercel

1. Push the project to a Git repository and import it in Vercel (framework: Next.js; defaults are fine).
2. In Project Settings → Environment Variables, add every variable you need from the table above. Set NEXTAUTH_URL to your production URL.
3. Create the database tables once, from your computer, with DATABASE_URL pointing at the production database:

npx prisma migrate deploy

4. Deploy. Vercel runs npm install, which triggers prisma generate, then next build.
5. Add the production redirect URI in Google Cloud and the production webhook in Stripe (sections above).

Scripts

Command	Purpose
npm run dev	Development server
npm run build / npm start	Production build / server
npm run lint	ESLint
npm run db:migrate	prisma migrate deploy
npm run db:push	prisma db push (prototyping only)
npm run db:studio	Prisma Studio

Project structure

app/            Routes: (marketing), (auth), (dashboard), api/
components/     UI components
features/       Billing, team, analytics logic
modules/        AI Writer service and UI
lib/            auth, db, openai, stripe, env helpers
prisma/         schema.prisma and migrations/
documentation/  Documentation.html (same content as this file)

Troubleshooting

• Build prints "[auth] Missing Google OAuth credentials": a warning only; set the Google variables to remove it.
• /writer redirects to /login: expected when signed out.
• Dev console shows "headers() should be awaited" errors: a known warning from next-auth 5.0.0-beta.19 in development only; the pages still work and it does not occur in production builds.
• "Timed out trying to acquire a postgres advisory lock": use the direct (non-pooler) database URL for prisma migrate, see "Database setup".
• P3005 The database schema is not empty: the database was created with db push; run the migrate resolve --applied command from "Database setup" once.
• Database errors on first load: run npx prisma migrate deploy and check DATABASE_URL.
• AI Writer shows "needs an API key" or returns 503: OPENAI_API_KEY is missing.
• AI Writer shows "Model unavailable": AI_MODEL is not a valid model at your provider; on OpenRouter check https://openrouter.ai/models.
• AI Writer shows "No credits" or "Rate limited": add credits at your provider or wait and retry. Free models are often rate limited; set AI_FALLBACK_MODELS.
• /billing shows "Demo UI — Connect Stripe Keys to Activate": STRIPE_SECRET_KEY is not set in this deployment; set it and redeploy.

License and credits

See LICENSE.txt and Credits.txt. You bring your own OpenAI, Stripe and database accounts; this product is not affiliated with them.