CRMaster: Deep NLP Integration Across the Entire ERP Workflow

In this work we present CRMaster, a prototype AI-augmented Customer Relationship Management (CRM) platform that integrates modern retrieval and machine learning techniques into a traditional sales pipeline. CRMaster is a Next.js–based web application featuring an AI-powered chat assistant, a knowledge-based search engine (using Retrieval- Augmented Generation, or RAG), dynamic lead scoring, customer segmentation, and analytics dashboards. The system is implemented with lightweight interpretable components: intent classification and AIML for conversational intent, a TF– IDF embedding for document encoding, a simple in-memory vector search for knowledge retrieval, an incremental linear model for lead scoring, and k-means clustering for segmenting customers. In demonstrations, CRMaster achieves real-time interaction and provides actionable insights. For example, retrieval-augmented answers based on company FAQs and product docs. We compare CRMaster’s design to existing CRM platforms and highlight its novel combination of RAG-style search with CRM workflows. Our prototype shows that integrating vector search and simple ML models into a CRM can accelerate sales response and forecasting. Future work will involve scaling out the vector database and integrating large language models for richer dialog.