amboras
Description: The Shopify of the AI Era — brand brief to live store in under 5 minutes
README
Amboras
Brand brief → live Shopify storefront in under 5 minutes.
Amboras lets a founder describe their product in plain English and receive a fully generated, Shopify-backed storefront — brand identity, copy, products, and checkout — with no design or development work.
---
Generation Flow
User types a brand brief
│
▼
┌───────────────────┐
│ /api/generate │ SSE stream opens — progress events sent to browser
│ │
│ 1. Call AI │──► OpenRouter → GPT-4o-mini
│ │ Structured tool call forces a typed StoreConfig:
│ │ brand, colors, hero copy, products, testimonials, FAQ
│ │
│ 2. Sync Shopify │──► Shopify Admin REST API (2025-01)
│ │ Creates real products + variants in the connected store
│ │ Returns variantIds used for cart checkout later
│ │
│ 3. Save store │──► File (.stores/id.json) locally
│ │ Upstash Redis (store:id key + store:index set) in prod
└───────────────────┘
│
▼
Dashboard /dashboard/[storeId]
│
├── Overview real order count + revenue from Shopify
├── Analytics daily revenue/orders chart, per-product sales (7/14/30d)
├── AI Editor describe a change → AI patches StoreConfig in place
└── Settings Shopify connection status, danger zone---
System Architecture
┌─────────────────────────────────────────────────────────────────┐
│ Browser │
│ │
│ / /store/[id] /dashboard/[storeId] │
│ Landing + Branded storefront Analytics, editor, │
│ brief form (client-rendered) settings │
│ │ │ │ │
└───────┼──────────────────┼──────────────────────┼───────────────┘
│ │ │
│ POST (SSE) │ GET /api/stores/[id] │ GET /api/shopify/stats
│ │ │ POST /api/stores/[id]/edit (SSE)
▼ ▼ ▼
┌─────────────────────────────────────────────────────────────────┐
│ Next.js API Routes │
│ (Node.js · Vercel serverless) │
│ │
│ /api/generate /api/stores/[id] /api/shopify/stats │
│ ───────────── ──────────────── ─────────────── │
│ Orchestrates AI GET → store JSON Fetch orders from │
│ + Shopify sync DELETE → remove Shopify, group by │
│ Returns SSE stream date + product │
│ │
│ /api/stores/[id]/edit /api/cart │
│ ────────────────── ────────── │
│ AI patches existing Build cart permalink │
│ StoreConfig via SSE → { checkoutUrl } │
└──────────┬──────────────────────────────────────┬──────────────┘
│ │
▼ ▼
┌─────────────────────┐ ┌────────────────────────────┐
│ OpenRouter │ │ Shopify Admin REST API │
│ → GPT-4o-mini │ │ │
│ │ │ POST /products │
│ Structured output │ │ GET /orders │
│ via tool_choice │ │ Cart permalink checkout │
│ forces typed JSON │ │ │
│ (no parsing risk) │ └────────────────────────────┘
└─────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ Persistence Layer │
│ │
│ Local dev Production │
│ ───────── ────────── │
│ .stores/[id].json Upstash Redis │
│ One file per store store:[id] → JSON blob │
│ store:index → ID set │
│ │
│ Same interface: getStore / saveStore / deleteStore │
│ lib/store-cache.ts detects env and routes accordingly │
└─────────────────────────────────────────────────────────────────┘---
Data Model
GeneratedStore
├── id string
├── brief BrandBrief (product, audience, vibe, competitors)
├── createdAt Date
├── editHistory[] { prompt, appliedAt }
│
├── config StoreConfig
│ ├── brand { name, tagline, colors ×4, fontStyle }
│ ├── hero { headline, subheadline, ctaText × 2 }
│ ├── products[] { name, description, price, features, emoji, imageQuery }
│ ├── stats[] { value, label }
│ ├── testimonials[]{ name, role, review, rating }
│ ├── about { title, body, values[] }
│ ├── faq[] { question, answer }
│ └── seo { title, description, keywords }
│
├── shopifyStoreUrl string (mystore.myshopify.com)
└── shopifyProducts[] { id, handle, title, variantId }
└── variantId used in cart permalink---
Key Design Decisions
| Decision | Why |
|---|---|
| SSE instead of WebSockets | Works with Vercel's streaming response model; no persistent connection needed |
| Single AI call with structured tool output | Forces a valid typed config — no free-form text to parse, no partial failures |
| Shopify as the commerce layer | Real inventory, real payments, real checkout; Amboras only handles storefront + AI |
| File storage locally, Redis in production | Zero infrastructure for development; same interface both ways |
| Client-side cart state | Avoids a cart DB entirely — Shopify checkout is built from variant IDs at checkout time |