How AI Powers the Next‑Gen Sugar BI Platform for Smarter Decision‑Making

Product Overview

Sugar BI, originally Baidu's internal ShowX reporting platform, was rebuilt in 2018 and rebranded. It now offers a SaaS BI service with over 30 data source connectors, zero‑code drag‑and‑drop visual analytics, and mobile‑responsive dashboards.

AI‑Driven Features

Since 2021 the platform has integrated large‑model capabilities, launching automatic analysis, variance attribution, and the Sugar Bot conversational assistant that translates natural‑language queries into SQL and visual charts.

BI Core Capabilities

Data source integration (relational, NoSQL, time‑series, Hive, Spark, etc.)

Semantic data model with automatic dimension/measure detection and virtual fields.

Interactive engine supporting filters, drill‑down, cross‑chart linking, and federated queries via Presto.

Rich visual library (150+ ECharts/D3 components, 3D charts, maps, media, themes).

Smart Chart Recommendation

The system extracts chart features, applies mandatory and optional scoring rules, and ranks suitable visualizations, automatically switching chart types as fields are added or removed.

Sugar Bot Architecture

Four‑step pipeline: prompt engineering, NL2JSON conversion, result correction, and smart‑chart rendering. Prompt includes schema, examples, output format, private domain knowledge, and user query. Few‑shot examples are generated automatically or retrieved from user‑feedback vectors.

Implementation Details

Key components: schema handling with semantic aliases, virtual computed fields, and geographic field tagging; few‑shot generation (cold‑start rules and feedback‑driven examples); private‑domain knowledge injection; and function‑call support for user‑specified chart types.