Challenges
1- Manual chatbot responses.
2- No automated escalation flow.
3- Fragmented data storage.
4- No real-time monitoring.
5- Scattered product content.
Solutions
1- RAG-powered responses.
2- Automated escalation emails.
3- Centralized Supabase storage.
4- FastAPI Admin Dashboard.
5- Website content scraper with Astra DB.
Results
1- Context-aware AI responses.
2- Automated escalation handling.
3- Real-time monitoring and reporting.
4- Instant CSV exports.
5- Reliable, domain-specific knowledge base.
The client, a tile and stone company, struggled to manage customer queries on product details, lead times, and issue reporting. Without a centralized system, responses were inconsistent, escalations were delayed, and there was no structured repository for conversation history.
They needed an intelligent chatbot that could provide accurate, context-based responses, automate escalations, and maintain a centralized record of all conversations. Real-time monitoring, analytics, and export capabilities were also essential.
An AI chatbot with RAG for contextual replies, automated escalations via Microsoft Graph API, centralized storage in Supabase, a FastAPI dashboard for monitoring/exports, and Astra DB for structured product knowledge.
- Inconsistent and Incomplete AI Responses: Chatbot answers often missed context or failed to resolve complex queries.
- Manual Escalation Workflow: Escalations were fragmented and time-consuming without automation.
- No Monitoring or Export Mechanism: Managers lacked real-time oversight of chatbot interactions.
- Lack of Audit-Ready Records: Conversation history wasn’t stored in a structured, retrievable format.
- Scattered Product Information: Critical details were spread across the website, limiting chatbot accuracy.
- Context-Aware Retrieval-Augmented Generation - RAG-based retrieval ensured contextual, accurate responses.
- Automated Escalation Flow - Microsoft Graph API automatically generated and sent escalation emails.
- Centralized Conversation Storage - Supabase stored structured conversations by session and timestamp.
- Admin Dashboard for Monitoring & Management - Admin Dashboard enabled monitoring, record filtering, and quick CSV exports for management use.
- Website Content Scraping & Storage - Product and policy data scraped and stored in Astra DB for chatbot retrieval.
95%
Instant, Contextual Chatbot Responses
The chatbot now delivers precise, context-aware answers in real time, minimizing repetitive queries and improving user engagement.
85%
Faster Escalation Handling
The automated escalation flow significantly reduced response time for complex cases, ensuring quicker resolutions and improved client satisfaction.
95%
Improved Data Accessibility
Centralized chat records enhanced compliance and visibility, with most conversations now stored in a structured and easily retrievable format.
88%
Expanded Knowledge Accuracy
Structuring website content in Astra DB increased the chatbot’s knowledge coverage, delivering more relevant responses and elevating customer satisfaction.
80%
Enhanced Reporting Efficiency
Managers can now generate and export reports within seconds, improving audit readiness and analysis efficiency.
