How AI Is Transforming Business Intelligence: From BI 1.0 to Sugar BI’s Intelligent Prediction

This article, originally presented at the DataFun Enhanced Analytics Forum in December 2022, explains how artificial intelligence is making Business Intelligence (BI) smarter, moving beyond historical analysis to provide future guidance.

1. BI Development History

The term “business intelligence” first appeared in 1865, followed by IBM scientist Hans in 1958 describing BI’s value, and Gartner’s formal definition in 1989. BI has progressed through three stages: BI 1.0, BI 2.0, and the emerging BI 2.5.

1.1 BI 1.0

Driven by data warehousing, BI 1.0 relied on static Excel reports generated manually by developers, resulting in hours‑long delays and costly, month‑long development cycles for visual dashboards.

1.2 BI 2.0

With the rise of internet technologies, self‑service analytics and agile BI tools allowed users of any skill level to create interactive dashboards quickly, though analysis quality varied and decision‑making remained limited.

1.3 BI 2.5 (Intelligent BI)

The current era adds AI‑driven capabilities, laying the foundation for intelligent BI.

2. Intelligent BI Era

AI+BI combines machine learning with traditional BI to create “augmented analytics,” a concept introduced by Gartner in 2017. It automates data preparation, insight discovery, and sharing, enabling anyone to obtain high‑quality analytics without deep ML knowledge.

Key intelligent features include:

Smart Charts : An algorithm recommends the most suitable chart type for a given dataset.

Smart Analysis : Automatic analysis, anomaly detection, and trend analysis are provided out‑of‑the‑box.

Smart Interaction : Voice‑enabled queries and multi‑device support turn the platform into a data assistant.

Smart Decision : Machine‑learning‑based predictive analysis assists leadership in strategic decisions.

3. Predictive‑Analysis Platform Landscape

Three main categories exist:

BI Platforms : Integrated predictive modules (e.g., Sugar BI DI) for non‑technical users; loosely coupled extensions for data scientists; or simple time‑series and clustering tools for everyday analysts.

Open‑Source ML Tools : Provide data‑preprocessing, model training, and evaluation, suitable for researchers but with higher adoption barriers in enterprises.

Integrated AI Development Platforms : Commercial solutions (e.g., Baidu BML) offering end‑to‑end model APIs for AI experts and data scientists.

4. Sugar BI’s Intelligent Prediction DI Module

The DI (Data Intelligence) module targets business users without ML backgrounds, offering a no‑code, fully integrated forecasting experience within the BI workflow.

4.1 Function & Value

DI helps users understand what is happening, why it happened, and what may happen next, turning data into actionable decision support.

4.2 Core Features

Built‑in Models : Supports clustering and linear regression for quick insights.

Automatic Model Selection : The system chooses the optimal predictive model without manual training.

Custom & Training Models : Users can train binary‑classification models or upload external models, with future plans for Open API integration.

4.3 Demonstrations

The demo shows that built‑in models create prediction fields instantly, while training models (e.g., binary classification) involve selecting data, configuring algorithms, and publishing the trained model for visualization.

5. Q&A Highlights

AI‑driven natural‑language query parses user utterances into data fields, enabling voice‑based chart generation. Sugar BI also offers a one‑month free trial and community support for further learning.