Generative BI Approach to Conversational Analytics

This document is a technical report that outlines the concept of Generative BI, which integrates Generative AI and Large Language Models (LLMs) into business intelligence and analytics processes. It describes a conversational analytics framework where users can interact with an AI to obtain insights from data without sending sensitive information to public services. The report details five key capabilities required for effective conversational analytics, including an AI strategy that works directly with data, a multi-LLM strategy, and the ability to provide sophisticated answers to complex questions. It also discusses the limitations of LLMs in handling quantitative data and presents a solution that involves using LLMs for natural language interpretation while performing analytic processing separately. The document emphasizes the importance of a semantic model that describes data structures, which allows LLMs to interpret user queries accurately. Additionally, it highlights the Pyramid Platform's capabilities in executing analytic recipes and generating dynamic visualizations based on user inquiries.