An AI Chat Agent That Sells, Supports, and Hands Off to Humans, Seamlessly

SandCloud is a mission-driven D2C lifestyle brand selling towels, apparel, and accessories through their Shopify storefront. As the brand scaled, their customer support volume grew with it, order status inquiries, product questions, returns, and shipping issues flooding in around the clock.

They needed a support solution that could handle the bulk of conversations autonomously while preserving the personal, brand-aligned experience their customers expected. Off-the-shelf chatbots felt generic and couldn't integrate deeply enough with their Shopify operations or their existing Kustomer support desk.

SandCloud's support team was stretched thin handling repetitive queries that didn't require human judgment, order tracking, product availability, and return policies.

But the brand couldn't afford a robotic chatbot experience that would erode customer trust. They needed an AI agent intelligent enough to resolve common issues end-to-end using live Shopify data, yet graceful enough to hand conversations to human agents when the situation called for it.

The solution also had to integrate with their existing Kustomer CRM without disrupting established support workflows, and deploy cleanly on their Shopify theme without introducing framework bloat or third-party script risks.

We approached this as an integration architecture problem, not just a chatbot build. The AI needed to sit between Shopify (for commerce data), Kustomer (for human support), and the storefront (for the customer experience), orchestrating all three without coupling them.

We chose n8n as the automation backbone because it gave us visual workflow design with webhook-native execution, letting us iterate on conversation logic without redeploying code. For the frontend, we deliberately avoided React or any framework. A single vanilla JavaScript class meant zero runtime dependencies, a sub-50KB payload, and no version conflicts with Shopify's theme stack.

Security was designed in from day one: Cloudflare Workers as auth proxies meant the n8n backend was never exposed to the browser.

The SandCloud Smart Chat Bot transformed customer support from a bottleneck into a scalable, always-on operation, handling routine queries autonomously while preserving human touch for complex issues.

• Fully automated AI chat support deployed on the Shopify storefront, handling order inquiries, product questions, and returns without human intervention.
• Seamless human handover to Kustomer agents with zero conversation context loss, customers never have to repeat themselves.
• Zero runtime dependencies and sub-50KB widget payload, ensuring no performance impact on storefront load times.
• Enterprise-grade security architecture: Cloudflare Workers as auth proxies, SRI asset integrity, cryptographic session IDs, and origin validation on every request.
• Path-aware deployment that intelligently shows the widget only where customers need help, product pages, collections, and hides it during checkout to avoid friction.
• Complete build-to-deploy pipeline producing deployment-ready artifacts with auto-generated SRI hashes and Shopify Liquid snippets.

Incredible Visibility delivered an AI support system that feels native to SandCloud's brand while running on a security-first, zero-dependency architecture built to scale.

This project demonstrates Incredible Visibility's ability to build production AI systems that integrate deeply with existing ecommerce and CRM infrastructure. Rather than deploying a generic chatbot, IV architected a solution that respects SandCloud's existing tool choices (Shopify, Kustomer) and enhances them with AI, proving that intelligent automation doesn't require ripping out established workflows. The zero-dependency frontend approach and Cloudflare Workers security layer reflect IV's emphasis on lean, secure, maintainable systems that D2C brands can operate without dedicated engineering teams.