Smart Online Boutique on GKE

AI-powered microservices commerce platform

Comment

Inspiration

E-commerce demos often highlight infrastructure but stop short of showcasing intelligence. We wanted to go beyond a static storefront and bring AI into the experience. Inspired by GKE’s 10th anniversary and the need for more AI-native commerce platforms, we reimagined the classic Online Boutique with personalization, reasoning, and interactivity.

Our vision was to create a platform that not only demonstrates scalability on Kubernetes, but also feels alive — making every customer interaction smarter.

What it does

The Smart Online Boutique is an AI-powered microservices commerce demo that combines the classic Online Boutique with two new capabilities:

🤖 AI Microservice (ai-svc): Provides personalized recommendations and intelligent responses via an OpenAPI-based backend.

📚 ADK Catalog Agent: Enables natural language queries and knowledge-driven catalog exploration, letting users discover products beyond simple filters.

Together, these components transform a static demo into a forward-looking AI-native commerce experience.

How we built it

We built on top of Google Cloud’s Online Boutique microservices demo, extending it with new AI-driven services:

Google Kubernetes Engine (GKE): Container orchestration for all services.

Docker + Kubernetes manifests: For containerization and deployment.

FastAPI + OpenAPI 3.0: Backend framework and service specification for the AI microservice.

ADK (Agent Development Kit): To implement agentic workflows for catalog reasoning.

Polyglot microservices stack: Go, Python, Java, Node.js — integrated seamlessly with the new AI services.

The system is fully automated through Kubernetes manifests under k8s/, with helper scripts for deploy/cleanup. Live services are deployed and accessible via GKE ingress.

Challenges we ran into

Service orchestration — Coordinating multiple polyglot services with new AI-driven microservices required careful integration.

OpenAPI alignment — Ensuring the AI microservice conformed to OpenAPI 3.0 spec for interoperability.

Agent reasoning — Designing catalog exploration workflows that feel natural yet performant.

Scalability testing — Validating everything on GKE while keeping deployments smooth.

Accomplishments that we're proud of

Successfully added an AI-native microservice alongside existing microservices.

Built a catalog agent that enables natural exploration of products.

Passed 100% validation checks (18/18 ✅).

Deployed a fully live, interactive demo — available with frontend, AI Swagger docs, and catalog agent UI.

Created a showcase for GKE at 10 years that highlights scalability + innovation.

What we learned

How to design AI microservices that integrate cleanly into polyglot environments.

The importance of agentic workflows for making catalogs smarter.

How GKE simplifies multi-service deployment at scale.

That combining traditional microservices + AI-native services creates a future-proof retail experience.

We also deepened our understanding of Kubernetes networking, observability, and service orchestration across heterogeneous workloads.

What’s next for Smart Online Boutique on GKE

🔗 Integrating Vertex AI / Gemini for advanced AI reasoning.

💳 Adding intelligent checkout and fraud detection.

📊 Enhancing observability with Prometheus + Grafana.

🌍 Scaling to multi-region deployments for global reach.

Our long-term vision: turn Smart Online Boutique into the go-to AI-native retail demo that inspires developers to build the future of cloud-native commerce.

✨ Built with ❤️ for the GKE Turns 10 Hackathon

Built With

Share this project:

Updates