Problem
Employees often struggle to quickly find accurate compliance information relevant to their region. An efficient, AI-driven solution is needed to provide instant, accurate answers to compliance-related questions.
Project Overview
iComply is an innovative chatbot developed during my tenure at Reckitt, designed to provide employees with quick and accurate answers to compliance-related questions via Microsoft Teams. As one of the first Retrieval-Augmented Generation (RAG) production implementations globally, iCompliance leverages advanced AI technologies to ensure employees have easy access to relevant corporate policies and guidelines.
Key Features
• Location-Based Policies: The chatbot dynamically adjusts its responses based on the user’s country, ensuring compliance answers are tailored to specific regional corporate policies.
• “Did You Mean?” Feature: This feature rejects questions that cannot be answered by existing corporate policies and suggests alternative questions that employees might ask, improving user experience and relevance.
• Hypotheses Evaluation: Multiple methods of document encoding, embeddings, and AI models were evaluated to achieve an impressive 96% accuracy in delivering pertinent answers to users’ inquiries.
Technical Highlights
iCompliance stands out for its sophisticated use of RAG technology and its integration into Microsoft Teams, making compliance information readily available to employees. The project involved extensive testing and optimization to ensure high accuracy and reliability.
