Integrations
| Source | Stack |
|---|---|
| Public | No single consumer URL — B2B distributor channels are private / messaging-based |
| Engineering | commercetools (B2B+B2C multi-store), WhatsApp + Meta Cloud API, LangGraph/LangChain/OpenAI, CSR order apps, Datadog, n8n |
Headless commerce meets social selling — B2B, B2C, and agentic AI
Rich Cookies USA needed more than storefronts — they needed social commerce where B2B distributors order from messaging apps, B2C headless experiences per brand, and custom tooling for customer service reps to create and modify B2B orders. I architect the composable commercetools stack across production applications today and the multi-tenant platform we're building next.
Production applications (live)
B2B + B2C multi-store
- commercetools Store concepts with channel-scoped pricing, inventory, and Business Units
- 5+ brand storefronts on Next.js + TypeScript with localized checkout
- JSON-LD structured data and Core Web Vitals tuning — Lighthouse 90+
Custom apps for service representatives
- CSR order console — create, modify, and manage B2B orders without Merchant Center limits
- Staged OrderEdits with approval flows tied to distributor Business Units
- Audit trails linking rep actions to Associates and orders in commercetools
Social commerce channel
Commerce-native conversations — every message ties back to Business Units, Associates, and Orders:
| Capability | What it delivers |
|---|---|
| Conversational ordering | Distributors check status, list orders, act on proposals from mobile |
| Shared CRM inbox | Sales sees associate, BU, linked order, full activity audit |
| Order change approvals | Interactive approve/reject on staged edits |
| Identity routing | Phone → verified Associate → scoped store, locale, currency |
Meta Cloud API + WhatsApp Business API with simulator mode for stakeholder demos.
Agentic layer — LangGraph + LLM (integrated with social commerce)
Phase 2 adds LLM-powered intent routing for free-text distributor messages — tool-calling against commercetools APIs:
- LangGraph.js state machines — plan → query → respond loops with visual flow charts for stakeholder review
- LangChain tool bindings for order lookup, proposal status, escalation
- OpenAI models with guardrails: scoped BU access, token budgets, human-in-the-loop for destructive actions
- Fallback to structured intents when confidence is low
Multi-tenant platform (in development)
- Shared storefront templates with per-tenant branding, catalog, and payment config
- Faster new-brand launch — target ~40% reduction in go-live time vs. one-off builds
- Datadog + n8n automation for sync monitoring and webhook orchestration
Technologies used
commercetools, Next.js, TypeScript, WhatsApp Business API, Meta Cloud API, LangGraph.js, LangChain, OpenAI, Node.js, Datadog, n8n, JSON-LD, Azure-ready deployment
Result
A unified social commerce + headless commerce platform: distributors order through messaging with LLM-assisted routing; consumers shop branded storefronts; CSRs manage B2B orders in custom apps — all on one composable commercetools foundation scaling across Rich's USA brands.
