noRoof Biz — White-Label E-Commerce & Business Management Platform

Context & The Business Problem

Small-to-medium businesses (SMBs) and local sports clubs in Uruguay often face steep barriers when establishing an online retail presence. Off-the-shelf platforms like Shopify or WooCommerce present specific challenges for this market:

1. Local Payment & Invoicing Friction: Lack of native, zero-friction integration with regional payment gateways (Mercado Pago) and local electronic invoicing standards.
2. Operational Complexity: Overly complex administrative panels that overwhelm non-technical store owners who just need to track stock, process orders, and send receipt notifications.
3. High Fixed Costs: Monthly subscription models in USD that cut heavily into small merchant margins.

The Mission: Architect a lightweight, multi-tenant platform that delivers high-performance Next.js storefronts for consumers and an intuitive Angular admin workspace for merchants—optimized specifically for regional commerce workflows.

Original Production Architecture

• Next.js storefront
• Angular admin
• Firebase
• Mercado Pago
• Brevo / Telegram

The original platform was engineered around a decoupled serverless architecture to guarantee isolation, low latency, and zero infrastructure maintenance costs during early scaling.

Architectural Decisions

• Frontend Dual-Framework Setup:

  • Next.js (Storefronts): Utilized for App Router Server Components, static site generation (SSG), and incremental static regeneration (ISR) to deliver near-instant loading speeds and optimal SEO indexing for merchant catalogs.
  • Angular (Merchant Dashboard): Chosen for its opinionated structure, robust reactive form handling (RxJS), and enterprise-grade state management for complex inventory tables and order management workflows.

• Backend & Multi-Tenancy (Firebase Stack):

  • Database: Cloud Firestore organized with strict multi-tenant collection security rules.
  • Auth & Security: Custom Firebase Authentication claims (store_admin, store_id) to enforce scope boundaries at the API layer, preventing cross-tenant data leaks.
  • Serverless Logic: Firebase Cloud Functions handling payment webhooks, transactional email dispatches via Brevo SDK, and instant order notification alerts sent to merchant Telegram groups.

The AI Evolution: Giving Merchants Their Time Back

• AI assistants

With the transactional flow working, the biggest remaining cost for merchants was the time they spent in the admin. That is where AI is going first.

Done: Two Features That Cut Admin Time

• Products from a Photo and a Text: Instead of filling in the whole product form, a merchant uploads an image and a short text, and the product and its items are drafted from them.
• Draft Orders from a WhatsApp Message: Many orders arrive by WhatsApp. Instead of picking each product by hand, the merchant pastes the message into the admin and a draft order is created from it.

Building Faster

AI-assisted, agentic coding workflows have made feature iteration around 100% faster, roughly twice the pace we had before, while keeping strict TypeScript validation with Zod schemas.

Roadmap: Features AI Makes Possible

These were out of reach for a platform this size before AI. None of them is built yet:

• Marketing Content & Copy Assistant: Help merchants write product copy and marketing content.
• AI Shopping Assistant: Guide customers from plain requests like “I want a gaming laptop to play GTA” or “I’m looking for a present for my wife” to the right products and suggestions.
• Support Agents (MCP): Let an agent answer questions like “Where is my order?” or “Has my payment cleared?” from the store’s own order data.

Key Outcomes & Engineering Learnings

Measureable Impact

• Zero Downtime Deployments: Successfully deployed live production storefronts serving active local merchants and sports club merchandise shops.
• Automated Customer Touchpoints: Replaced manual receipt emailing with 100% automated transactional messaging across Telegram and Brevo.
• Faster Catalog Onboarding: Products are created from a photo and a text instead of a full form.
• Orders Straight from WhatsApp: Pasting a customer’s message replaces picking products one by one.
• Twice the Iteration Speed: Feature iteration is around 100% faster with AI-assisted development.

Key Engineering Takeaways

1. AI Drafts, Merchants Confirm: Both AI features produce drafts, a filled-in product and a draft order, so they save time without deciding stock or orders on their own.
2. Multi-Tenant Security First: Designing custom claims in Firebase early saved hundreds of hours of refactoring when scaling from a single store to a multi-tenant platform.
3. Build Around Existing Habits: Merchants already receive orders on WhatsApp. Meeting them there removes a step instead of asking them to change how they work.
4. Pragmatic Technology Choices: Choosing the right tool for the job (Next.js for consumer web vs. Angular for complex administrative tools) yields better user experiences than trying to force a single framework across all domains.