Usage Guide

This document is not organized by job role, but by what you need to do right now. The same person may cover both zones at different times — asking questions and checking data today, adding tables and adjusting semantics tomorrow — this is the norm, not the exception.

Analytics Agent capabilities fall into two zones:

Zone	What You Do	What You Need to Know
Analyze Data	Ask questions, read answers, dashboards, explore, feedback	What data is in the analytics domain, how to ask better questions
Administer & Maintain	Add tables, configure fields, build metrics, set permissions, check audits	Data model, SQL syntax, permission system

These two zones are not independent. The simpler and more natural the business users' experience, the more thoroughly the administration side needs to nail domain boundaries, field semantics, metric definitions, knowledge, answer builders, permissions, and validation. Without this backend foundation, a natural language frontend won't consistently produce stable, accurate answers.

Think of administration and maintenance as the process of "harnessing a large model into an enterprise analysis Agent": not letting the model directly guess at enterprise data, but giving it explicit business context, available resources, field meanings, metric definitions, permission boundaries, and validation feedback. Only then can business users rely less on underlying technology and more on natural language to complete analysis.

---

Analyze Data

In the Analyze Data zone, you use natural language to get answers, explore metrics, and build and use dashboards. You don't need to know underlying table names or SQL.

Features You'll Use

Feature	What It Does	Entry Point
Analytics Q&A	Look up numbers, view trends, make comparisons using natural language	Analytics > Select Domain > Analytics
Data Tab	View what tables and fields the current domain has, confirm what can be asked	Analytics > Select Domain > Data
Explore	Perform benchmark/comparative time analysis and drill-down around configured metrics	Analytics > Select Domain > Explore
Recommended Questions	Click example questions when you don't know what to ask	Analytics > Popular Question Search
SQL Statement	For BI analysts or maintainers to verify what fields and conditions important answers used	Q&A Result > SQL Statement
Logs	For BI analysts or maintainers to see why the system responded this way and which configurations were matched	Q&A Result > Logs
Save to Dashboard	Preserve valuable charts to a dashboard	Q&A Result > Chart top-right > Save to Dashboard
My Dashboards / Shared Dashboards	Maintain or view team dashboards	Dashboards
Scheduled Tasks	Periodically push analysis results or anomaly monitoring	Scheduled Tasks
Table Rendering	Adjust table style and layout	Q&A Result > Table Rendering
Dashboard Versions	View or revert dashboard change history	Dashboard Detail > Version History
Chart Refresh	Set automatic dashboard data refresh frequency	Dashboard Detail > Refresh Configuration
Feedback	Submit correction feedback when answers are wrong or charts are incorrect	Q&A Result > Feedback

Typical Path

Select the right analytics domain → Open "Data" tab or check recommended questions to understand what can be asked → Ask questions in natural language → View answers, charts, and tables → Submit important results to BI analyst or maintainer to verify SQL and logs → Stabilize analysis into dashboards → Configure scheduled tasks for recurring questions → Submit feedback when answers are wrong

Boundaries of the Analyze Data Zone

When you encounter the following, you have entered the Administer & Maintain zone:

• Looking at recommended questions and finding the domain is missing tables or fields → Need to add tables or configure field semantics
• Explore page doesn't show needed metrics → Need to create metrics or answer builders
• The system consistently misunderstands a business term → Need to configure knowledge or field aliases
• Users can't see data they should see → Need to check permissions or row-level permissions

You don't have to do these yourself — but if you're a one-person team, see the next section.

---

Administer & Maintain

In the Administer & Maintain zone, you enable the system to correctly understand data, manage who can see what, and track who changed what.

The delivery goal of administration and maintenance is not "fill in all configuration items" — it's to let business users get answers that match business definitions and permission boundaries with less need to understand underlying fields, write technical conditions, or judge SQL.

Features You'll Use

Semantic Layer

Feature	What It Does	Entry Point
Table Management	View and add tables available for Q&A	Data > Tables
Field Semantics	Configure aliases, descriptions, types, usage, hidden status	Data > Tables > Field Configuration
Virtual Columns	Create derived fields using expressions	Data > Tables > Virtual Columns
Metrics	Solidify aggregation definitions	Data > Metrics
Answer Builder	Encapsulate complex SQL templates, support dynamic dimensions and filters	Data > Answer Builder
Knowledge	Supplement business terminology, definitions, and synonyms	Data > Knowledge
Files	Upload documents for document Q&A	Data > Files
Index Management	Manage field retrieval indexes	Data > Index Management
Recommended Questions	Design guidance questions for the analytics domain	Domain > Basic Information > Recommended Questions

Governance

Feature	What It Does	Entry Point
User Management	Manage accounts, user groups	Administration > User Management
Role Authorization	Create roles, configure feature permissions and resource permissions	Administration > User Management > Authorization Management
Row-level Permissions	Control users to see only designated data rows	Administration > User Management > Row-level Permissions
Analytics Domain Permissions	Control who can enter which domain	Domain Configuration > Permissions
Audit Log	Track who did what at what time	Administration > Audit Log
Feedback Management	Centrally view and assign user feedback	Administration > Feedback
Data Sources	Manage underlying data connections	Administration > Data Sources
Model Configuration	Manage available LLM models	Administration > Model Configuration
Scheduled Task Governance	View task status, send addresses, disable/enable	Scheduled Tasks

Validation

Feature	What It Does	Entry Point
Q&A Validation	Test configurations using typical questions	Analytics > Ask Question → Check SQL and Logs
Health Score Scan	Check for missing field descriptions, metrics, and table relationship anomalies	Domain > Basic Information > Health Score

Typical Path

The person analyzing data tells you: can't find a certain type of question, answers are wrong, definitions are off → If it's a table/field problem: add table → configure field semantics → hide sensitive fields → If it's a definition problem: create metrics or answer builders → configure knowledge → If it's a permission problem: add to domain → configure role → configure row-level permissions → Validate with typical questions → verify SQL and logs → update recommended questions → If the problem recurs: go back to analytics domain planning, consider splitting the domain

This path needs to form a closed loop: business users expose problems through natural language, BI analysts convert problems into verifiable cases, semantic layer maintainers fix fields, metrics, knowledge, or answer builders, administrators check permissions and audits, then the same business questions are used to validate the results. Without this loop, simply lowering the barrier for business users to ask questions won't make answers more accurate.

---

Relationship Between the Two Zones

Dashboards

Dashboards are analytical assets, not administrative operations.

Zone	What You Do
Analyze Data	Create dashboards, add charts, organize content, share with team
Administer & Maintain	Adjust permission boundaries when needed, track dashboard changes through audits

Scheduled Tasks

Scheduled tasks span both zones — anyone can create them, but governance needs to manage them.

Zone	What You Do
Analyze Data	Design monitoring scenarios, create tasks, edit execution time and content
Administer & Maintain	Govern send addresses, disable abnormal tasks, check execution permissions

Feedback

Feedback is the closed-loop entry point between both zones.

Zone	What You Do
Analyze Data	Submit "what's wrong" and the correct business understanding
Administer & Maintain	View, assign, fix, and re-validate in Administration > Feedback

Column Hiding

Column hiding is a semantic configuration that achieves column-level governance — it belongs to the Administer & Maintain zone.

---

Reading Path

Starting from Analyze Data

1. Question Asking Guide — Structure of good questions
2. Analysis Patterns Guide — Looking up numbers, comparisons, trends, rankings, attribution
3. Reading Analysis Results — Reading values, tables, charts, SQL, logs
4. Using Dashboards — From Q&A to dashboards
5. Using Data & Exploration — Data tab + structured exploration
6. Handling Feedback — Submitting and improving
7. Notifications — Viewing background task status
8. Change Logo, Theme, and Chart Colors — Account info, Logo, theme, and chart colors

Starting from Administer & Maintain

1. Analytics Domain Planning Guide — Plan domains first, then add data
2. Configure Analytics Domain — Create domain, add tables, configure table details
3. Configure Field Semantics — Aliases, descriptions, types, usage
4. Configure Virtual Columns — Derived fields
5. Answer Builder Best Practices — SQL template design
6. Configure Knowledge — Definitions and terminology
7. Troubleshooting Q&A Accuracy Issues — Diagnosis and fixes
8. Governance Overview — Permissions, audits, domain isolation
9. Permission Management — Accounts, roles, domain permissions
10. View Audit Logs — Operation record tracking
11. Full Download and Data Export Governance — Download permissions and export auditing

Validation Loop After Configuration

Configure or modify → Ask with typical questions → Check SQL and logs → Pass → Hand off to data analysts → Fail → Go back to previous step

---

Core Principles

• Act based on capability, not job title labels. The same person can both analyze data and manage the system.
• Dashboards are analytical assets; the person asking questions and building dashboards manages content; permissions and audits belong to Administer & Maintain.
• Feedback is the closed loop between both zones — data analyzers submit, maintainers fix.
• Analytics domains are governance boundaries. Don't mix unrelated business in one domain; don't put all permission governance pressure on row-level permissions.

Related Documentation

• Quick Start — Run through your first Q&A in 5 minutes
• Analyst Guide — Complete documentation directory for the Analyze Data zone
• Configuration Guide — Complete documentation directory for the Administer & Maintain zone
• Governance Overview — Permissions, audits, domain isolation system
• Notifications — Background task status viewing
• Change Logo, Theme, and Chart Colors — Settings entry in the user avatar menu