AI-Powered e-commerce Enhancement – Jade Business Services LLC

Case Study: AI-Powered E-commerce Enhancement

Customer: Large Manufacturing Company

Problem Statement

The client, a large manufacturing company, faced significant challenges in their e-commerce platform’s sales and support channels, impacting both B2B and B2C segments. The existing process relied heavily on traditional, keyword-based search and manual human support, leading to considerable inefficiencies. Earlier Process Flow: Customers visiting the website would use a traditional search method, often relying on keyword searches. This frequently resulted in inefficient product discovery, particularly for complex refrigeration systems and highly specific parts. For technical inquiries, installation guidance, or precise part identification, customers were often funneled towards costly human support.

This manual approach for frequent, yet solvable, issues consumed valuable resources and often led to customer frustration and delays due to inconsistencies and long wait times. If customers needed “DIY help” or were not “Ready to Buy,” they often required direct human intervention. Cart abandonment was common, with re-engagement efforts often failing.

Solution Section

Impact

These operational inefficiencies directly translated into significant negative consequences for the business:

Missed Sales Opportunities:
Ineffective customer interactions and a cumbersome product discovery process led directly to lost revenue.

Project Delays for Businesses:
B2B clients experienced substantial setbacks as they struggled to quickly obtain necessary information or identify specific parts, impacting their project timelines.

Sub-optimal Customer Journey:
The overall customer experience was characterized by frustration and perceived inefficiency, leading to reduced satisfaction and potential loss of repeat business.

Solution Delivered

To address these challenges, a Retrieval-Augmented Generation (RAG)-based Generative AI (GenAI) Assistant was deployed directly within the e-commerce platform. This comprehensive solution not only transformed customer interactions by delivering accurate, context-aware responses in real time, but also enhanced internal support processes by providing employees with instant access to relevant knowledge, reducing response times and improving overall efficiency. By combining advanced retrieval capabilities with generative AI, the assistant ensured that both customers and support staff received precise, up-to-date information, ultimately driving higher satisfaction and operational excellence.

Changed Process Flow

The AI-powered search method now uses Natural Language Processing (NLP) to provide highly intelligent and relevant search results, greatly enhancing product discovery. When customers need “DIY help,” the AI Assistant provides immediate, step-by-step instructions through an “AI Manual Search.” This significantly reduces the need for direct human support for routine queries. For customers ready to buy, the path to “Add to Cart” and “Checkout” is smoother. Critically, if a cart is abandoned, the GenAI Assistant proactively re-engages customers with personalized prompts and discount offers, leading to higher success rates in conversions. Human support agents are now augmented with AI tools, allowing them to focus on complex, high-value customer issues, vastly optimizing resource allocation.

Solution Architecture Overview

The system ingests various data sources like PDFs, text, Excel, CSV, and JSON from Cloud Storage. These documents are processed through PDF Extractor and Gemini OCR, then synthesized. Structured data undergoes Vector Embedding in BigQuery. This consolidated data feeds into Admin, Staff, and Public Databases. On the frontend, a User Interface handles User Queries which are routed through various agents, utilizing User Access Authentication (Cloud IAM). These agents interact with backend APIs, which in turn access the respective databases. The core GenAI functionality, powered by Gemini, aggregates data and dynamically prompts the Answer Agent to deliver final answers and sources to the user. The application deployment pipeline utilizes Code Repo, Docker, Artifact Registry, Cloud Run, Cloud Deploy, and Cloud Endpoints.

Business Benefits Section

Business Benefits

The implementation of the RAG-based GenAI Assistant yielded immediate and impactful business benefits:

Significant Improvement in Search Efficiency and Accuracy:
Achieved over 92% search accuracy, with query responses delivered in less than 8 seconds, drastically streamlining the product discovery process.

Reduced Cart Abandonment and Increased Conversions
A notable 17% drop in cart abandonment was observed among pilot customers, directly translating into increased sales.

Enhanced Customer Experience and Engagement:
The improved self-service "DIY" experience and enhanced product understanding fostered greater customer satisfaction and a more efficient, engaging shopping journey.

Technology Stack