FlagshipLead Developer2026 — Present

Unified Commerce Chat — AI-Powered Social Selling

One inbox for Facebook, Instagram & WhatsApp with AI, inventory, and orders.

Lead Full Stack EngineerInternal product at Yeamazing
Illustration representing Unified Commerce Chat — AI-Powered Social Selling

The problem

Yeamazing sellers were juggling customer conversations across Facebook comments, Instagram DMs, and WhatsApp chats while tracking orders and stock in separate spreadsheets. The goal was one product where every customer message, product, and order lives together — and an AI assistant handles repetitive selling tasks.

A social commerce command center where sellers connect their Facebook, Instagram, and WhatsApp pages to a single inbox, manage products and inventory, handle orders, and let an AI assistant answer customers, check stock, create orders, and share order status.

What I built

  • Unified social inbox

    Messages from Facebook, Instagram, and WhatsApp converge into one thread per customer, no matter which channel they use.

  • AI sales assistant

    The assistant answers product questions, checks live inventory, creates orders, and shares order status without leaving the chat.

  • Product & inventory

    Sellers add products, manage stock, and set variants; the AI and order flow always see current availability.

  • Order management

    Create, edit, and track orders from the chat or a dedicated dashboard with payment and fulfillment status.

  • Meta page connection

    OAuth-based Facebook/Instagram page and WhatsApp Business account connection with scoped permissions.

  • Real-time sync

    Webhooks from Meta plus Socket.io keep the inbox, inventory, and AI state in sync across devices.

Hard parts

  1. 01

    Designing a cross-channel conversation model

    Each platform has different message shapes, rate limits, and identifiers. I designed a normalized conversation schema and adapter layer so the UI treats every channel the same way while preserving platform-specific metadata.

  2. 02

    Meta app setup, permissions, and webhooks

    Reading the Meta docs, choosing the right app type, requesting permissions, verifying webhooks, and handling subscription approvals was the steepest part. I built a repeatable setup guide and fallback polling for cases where webhooks are delayed.

  3. 03

    Giving AI access to live business data

    The assistant needs to know stock, prices, and order status in real time. I exposed controlled tool calls over the product and order APIs so the AI can act without risking inventory or customer data.

My contribution

  • Leading end-to-end product design and technical architecture for the unified commerce platform.
  • Designed the chat-to-order workflow so conversations from Facebook, Instagram, and WhatsApp land in one inbox with full customer context.
  • Built product and inventory management modules that feed real-time stock data into the AI assistant and order flow.
  • Integrated AI assistant that can answer product questions, check inventory, create orders, and report order status from inside the chat.
  • Implemented Meta app setup, page connection flow, webhook ingestion, and socket-based real-time message sync.
  • Owned order management from cart creation through payment and fulfillment tracking.

Outcome

  • Sellers can manage all social sales channels from one dashboard.
  • AI handles routine product and order questions, reducing manual response load.
  • Orders are created directly from chat threads, shortening the sales cycle.
  • Next.js
  • TypeScript
  • Node.js
  • Socket.io
  • PostgreSQL
  • Meta Graph API
  • OpenAI

More case studies