How Harness + Skill Enable a New ChatBI Paradigm

Problem

Traditional data products suffer from inconsistent metric definitions, difficulty extracting data, and manual attribution because earlier AI lacked sufficient semantic understanding and reasoning.

ChatBI Solution

ChatBI implements a Multi‑Agent collaboration framework where each sub‑agent has a dedicated responsibility. A memory‑management layer separates short‑term conversational context from long‑term user preferences. Tool orchestration is performed by Harness, and a Skill management layer standardizes data‑retrieval and attribution capabilities.

Technology Selection

The team compared the Claude SDK with Ali Agent Scope, evaluating integration effort and performance, and selected the option that best satisfied integration and latency requirements.

Core Components

Metric engine that indexes business metrics and supports fast lookup.

Topic‑governance system that organizes knowledge documents and governs attribution logic.

Concrete methods for metric identification and dimension recognition as described in the solution documentation.

Quality Assurance

Accuracy evaluation using a test set to measure correct metric retrieval.

Automated test suites that validate agent responses and tool integrations.

Active clarification mechanism that prompts users when ambiguity is detected.

Link‑funnel visual monitoring to trace data flow and detect failures.

Confidence grading that assigns trust levels to each data result.