Olea — Modern Mediterranean

A fictional restaurant website: a sleek promo site for Olea plus a table assistant that recommends dishes based on a guest's preferences.

A single Next.js app (App Router). The menu is static data; the assistant runs on a deterministic rule-based engine and upgrades to an LLM (Google Gemini) when an API key is provided. The Gemini key stays server-side in a Route Handler, so nothing sensitive reaches the browser — and the whole thing deploys to Vercel as one project.

Stack

• Next.js 15 (App Router), TypeScript, Tailwind CSS v4, Framer Motion
• AI: Google Gemini (AI Studio free tier) — optional, key in env
• pnpm workspaces monorepo, local dev in Docker

Structure

apps/
  web/                 # Next.js app (UI + assistant route handler)
packages/
  menu-data/           # shared types, menu data, recommendation engine

packages/menu-data is the single source of truth: its types, menu data, and rule-based recommendDishes engine are all consumed directly by the app — no database needed (the menu is static).

• Server Components read the menu in-process via apps/web/src/lib/menu.ts.
• The assistant (client) calls POST /api/assistant/recommend — a Route Handler (apps/web/src/app/api/assistant/recommend/route.ts) that runs Gemini when a key is set, otherwise the rule-based engine. Key never leaves the server.

Develop

docker compose up        # dev server on http://localhost:3000

Optional: enable the LLM locally by creating apps/web/.env (gitignored) with GEMINI_API_KEY=... — see apps/web/.env.example. Without it, the assistant uses the rule-based engine.

docker compose exec web pnpm --filter @olea/web build   # production build
docker compose exec web pnpm --filter @olea/web lint

Deploy (Vercel)

1. Import the repo in Vercel.
2. Root Directory: apps/web (Vercel detects the pnpm workspace and installs from the repo root automatically).
3. Environment Variables: add GEMINI_API_KEY (and optionally GEMINI_MODEL). Leave them out to ship the rule-based assistant.
4. Deploy. Framework preset is auto-detected as Next.js.

Pages

• / — home (hero, signature dishes, story, hours)
• /menu — full menu with dietary & spice filters
• /about — the story
• /contact — details, Google Maps embed, reservation form (mock submit)

The table assistant is the floating button on every page.