End-to-end Shopify integration, abandoned cart recovery pipelines, and AI-powered payment validation — deployed for BlockBag Colombia.
BlockBag operated with a predominantly manual, WhatsApp-mediated sales process creating systemic revenue leakage at every conversion touchpoint — with no structured abandoned cart recovery and no programmatic data capture in Shopify.
Manual order capture resulted in untracked sales and inventory discrepancies within Shopify, limiting downstream analytics.
No automated recovery mechanism for customers who initiated checkout but did not complete payment — revenue permanently lost.
Human agents required at every stage of the sales funnel, creating throughput bottlenecks and inconsistent customer experiences.
Dual-cycle payment routing (Quincena 1 / Quincena 2) enforced manually, causing frequent reconciliation errors.
Customer PII existed in WhatsApp threads rather than structured Shopify records, blocking CRM and reporting capabilities.
Receipt screenshots reviewed manually by the account owner, creating latency and subjective error risk in payment confirmation.
A distributed, event-driven platform composed of four tightly integrated layers — each independently scalable and designed as a replicable module for future client deployments.
checkout.abandoned webhooks to Make.com scenarios that trigger re-engagement WhatsApp messages via the bot's authenticated API endpoint.Complete autonomous customer journey from first contact to confirmed Shopify order — zero human touchpoints required.
Seven independently deployable modules, each decoupled from the core flow to enable client-specific customization without re-architecture.
| Component | Technology | Role | Rationale |
|---|---|---|---|
| Runtime | Node.js 18 LTS | Bot server / API gateway | Non-blocking I/O for high-concurrency WhatsApp sessions |
| WA Bridge | Baileys | WhatsApp session mgmt. | Cost-efficient approach prior to Meta Cloud API approval |
| Hosting | Railway | Deployment & scaling | Zero-configuration CI/CD from GitHub with environment isolation |
| E-Commerce | Shopify REST API | Product & order data | Source of truth for catalog, COP pricing, and order records |
| Automation | Make.com (EU1) | Workflow orchestration | Low-latency event handling for abandoned cart recovery pipelines |
| Edge Compute | Cloudflare Workers | API proxy / CORS bridge | Zero-infrastructure overhead for serverless AI validation routing |
| CDN | Cloudflare Pages | Demo deployment | Global CDN with sub-100ms TTFB for static assets |
| AI Vision | Anthropic Claude Vision | Receipt classification | High-accuracy payment document classification at the edge |
| Version Control | GitHub | Source management | Continuous deployment integration with Railway via webhooks |
All sensitive data flows were engineered to minimize surface area, eliminate unnecessary persistence, and enforce access controls under strict client data protection agreements.
By eliminating manual touchpoints at each stage of the sales funnel, the platform enabled BlockBag to operate a high-throughput conversational commerce channel with zero incremental staffing cost.
| Impact Area | Before | After | Delta |
|---|---|---|---|
| Order Registration | Manual (human entry) | Automatic via Shopify API | ~100% effort reduction |
| Cart Recovery | None (zero re-engagement) | Automated Make.com pipeline | New revenue channel |
| Payment Routing | Manual cycle tracking | Programmatic bi-cycle logic | Zero reconciliation errors |
| Receipt Validation | Manual screenshot review | Claude Vision AI classification | Overhead eliminated |
| Owner Notifications | All messages forwarded | Selective high-signal events | Alert fatigue eliminated |
| Response Time | Minutes to hours (human) | <2 seconds (automated) | >99% latency reduction |
| Data Capture | Fragmented WhatsApp threads | Structured Shopify records | Full analytics unlocked |