Intelligent CRM System with AI-powered Assistant- Telkom

🗃️ Intelligent CRM System with AI-Powered Assistant

Overview

This project is a comprehensive Customer Relationship Management (CRM) system, designed to streamline telecom operations with advanced features for customer management, service lifecycle tracking, document automation, and intelligent analytics.
It integrates an AI-powered assistant that leverages a RAG (Retrieval-Augmented Generation) pipeline to allow users to interact with structured and unstructured data using natural language queries.

Built with scalability, usability, and data privacy in mind, this system improves operational efficiency and decision-making for telecom account managers and analysts.

🧰 Key Features

• 📇 Customer & Supplier Management:
  • Create, edit, and track customers and third-party suppliers.
  • Role-based permissions to ensure secure data access.
• 🔄 Service Lifecycle Management:
  • Activation, upgrade, downgrade, suspension, restoration, and cessation of telecom services.
  • Automated versioning and service ID generation.
  • Cost tracking for on-net and third-party services.
• 🧾 Document Automation:
  • Automated generation and storage of Service Order Forms (SOFs) and Handover documents as downloadable PDFs.
  • Support for uploading and managing signed documents.
• 📊 Billing & Revenue Tracking:
  • Prebilling logic with backbilling and final billing calculations.
  • Billing dashboards with KPIs: revenue trends, churned and suspended MRC, YTD revenue.
• 🤖 AI Assistant:
  • Natural language interface for querying structured CRM data (PostgreSQL) and unstructured documents (SOFs, Handovers).
  • Uses RAG pipeline with Weaviate and LangGraph for accurate, context-aware responses.
  • Self-healing SQL query generation with retries and clarifications.
• 📈 Dynamic Dashboards & Reports:
  • Interactive analytics dashboards with date filtering, animated charts, and downloadable reports.

🔷 Approach & Techniques

• Backend:
  • Built with Django, Django REST Framework, PostgreSQL for structured data.
  • Document storage and signals to auto-index into vector database.
• AI & NLP:
  • RAG pipeline combining:
    • Weaviate (self-hosted vector database) for document embeddings and retrieval.
    • LangGraph agentic framework for orchestrating workflows, retries, and summarization.
    • LLMs (local or cloud) for SQL generation and document summarization.
  • Embedding and chunking pipelines for document ingestion.
• Frontend:
  • Responsive UI with Bootstrap and AJAX-powered dynamic components.
  • Assistant chat modal embedded within CRM interface.
• Roles & Security:
  • Multi-level role-based access control for admin, account managers, analysts, and directors.
  • Permission checks integrated into AI assistant to enforce data visibility rules.

🚀 Tech Stack

• Language & Frameworks:
  • Python, Django, Django REST Framework
  • PostgreSQL
  • Bootstrap, jQuery/AJAX
• AI/NLP:
  • LangGraph
  • Weaviate (self-hosted)
  • Large Language Model (OpenAI GPT or local LLM)
• Other Tools:
  • ReportLab for PDF generation
  • FAISS (optional alternative for vector store)
  • Docker for local Weaviate deployment

📊 Impact

This project demonstrates how an agentic AI workflow combined with a robust CRM platform can enable business teams to make faster, smarter decisions by: - Automating repetitive operations. - Enabling intuitive data access through natural language. - Reducing error rates in reporting and billing. - Enhancing customer experience with timely service delivery and transparency.

🌟 Future Enhancements

• Integration with external APIs (e.g., SMS/email notifications).
• Advanced anomaly detection for fraud prevention.
• More granular analytics and predictive models for churn and upsell opportunities.