| Method | Process | Time |
|---|---|---|
| Manual | Write SQL → execute → Excel chart → slide | 30 min ~ hours |
| BI Dashboard | Natural language input → AI auto-generates chart | 30 sec |
Access Path
Access via Admin > Monitoring > Dashboard.| Permission | Dashboard List | Create/Edit/Delete | Shared Dashboards |
|---|---|---|---|
| Admin (monitoring write) | All | ✓ | ✓ |
| Admin (monitoring read) | All | ✗ | ✓ |
| Regular user | Owned + public/shared only | ✗ | ✓ |
Creating Dashboards
Create dashboard
Click Create Dashboard in the dashboard list.
| Field | Required | Description |
|---|---|---|
| Name | ✓ | Dashboard name (must be unique system-wide) |
| Description | ✗ | Dashboard purpose |
Add panels
From the created dashboard, add panels via Add Panel or AI Assistant. See Panel Composition and AI Assistant below.
Panel Composition
A dashboard consists of individual panels. Each panel has its own DB connection and SQL.Panel Types
- Chart
- Card
Visualize SQL execution results as various charts. Enter a natural language question in the panel editor and AI auto-generates SQL.Supported chart types:
Click the chart-type icon at the top of the panel in edit mode for instant switching.
| Chart | Description | Use Case |
|---|---|---|
| Table | Display data in table form | Detailed data viewing |
| Bar | Per-item comparison | Per-model usage count |
| Grouped Bar | Compare multiple grouped series | Per-model · per-period cost comparison |
| Line | Trend over time | Daily token usage trend |
| Area | Trend + area emphasis | Cumulative usage change |
| Pie | Ratio distribution | Per-agent usage share |
| Scatter | Correlation between two variables | Response time vs. token count |
| Histogram | Numeric distribution | Response time distribution |
| Heatmap | 2D matrix value distribution | Per-time-of-day · per-day-of-week usage |
Panel Editor
Click Add Panel or the edit icon on an existing panel to open the panel edit modal.| Field | Description |
|---|---|
| Panel name | Panel title (required) |
| Database | Pick the DbSphere to connect |
| Model | AI model used for SQL generation |
| Natural language query | Describe the data you want in natural language |
| Generate SQL | AI auto-generates SQL + shows explanation |
| SQL editing | Edit the generated SQL directly |
| Execute | Run SQL and show chart preview |
| Period filter | Specify date column to integrate with the dashboard period filter |
The generated SQL can be edited freely. When the AI-generated SQL isn’t accurate, edit directly and click Execute.
AI Assistant
Click the purple AI Assistant button at the bottom-right of the dashboard to open the chat panel.Conversational Dashboard Builder
Add, edit, and delete panels by chatting in natural language.Request in natural language
Examples:
- “Make a chart of this month’s per-model token usage”
- “Add a per-user cost card too”
- “Change the period to 30 days”
- “Change the pie chart to a bar chart”
Review AI results
The AI generates SQL and auto-places panels. Positions are auto-adjusted to avoid overlap.
One-shot Generation
Starting fresh with an empty dashboard, describe the dashboard’s purpose to the AI to generate a suitable panel set at once. DB connection is auto-tested first, generating only when valid.Filtering
The dashboard top has period filter and custom filters.Period Filter (TimeRangePicker)
| Preset | Description |
|---|---|
| Today / Yesterday | Today / Yesterday |
| This Week / Last Week | This week / Last week |
| This Month / Last Month | This month / Last month |
| 7d / 30d | Last 7 / 30 days |
| All | Entire period |
| Custom | Specify start/end dates directly |
period filter enabled.
Custom Filters (FilterBar)
In edit mode, use the Add Filter button to add filters that apply to the entire dashboard.| Filter Type | Description |
|---|---|
| Text | Free text input (e.g., username) |
| Date Range | Start ~ end date range |
| Select | Selection options |
Sharing
Share completed dashboards with team members.Set access scope
| Access Scope | Description |
|---|---|
| Public | All signed-in users can access |
| Restricted | Only specified groups or organizational units can access |
| Private | Only owner and Admin can access |
HTML Export
Export dashboards as standalone HTML files. Click the Export button at the top of the dashboard to execute SQL with the currently applied period and filter conditions, downloading the HTML file with the latest data. Exported file characteristics:| Item | Description |
|---|---|
| Format | Single HTML file (no external dependencies) |
| Chart interaction | Hover, zoom, and other Plotly default features supported |
| Theme | Light/dark mode auto-supported (follows system setting) |
| Layout | Same 12-column grid as the dashboard |
| Responsive | Charts auto-adjust on window resize |
Security
| Item | Policy |
|---|---|
| SQL execution | Only SELECT queries allowed. INSERT/UPDATE/DELETE/DROP and other modification queries are blocked |
| Filter input | Special character escaping to prevent SQL Injection |
| Share access | Access control (Public/Restricted/Private)-based verification |
| DB connection | Only read-only connections registered in DbSphere |
Use Cases
Executive KPI Monitoring
Executive KPI Monitoring
- AI auto-generation via “Make a key KPI dashboard for this month”
- Card panels show key values (total revenue, active user count, etc.)
- Line charts show daily trends
- Share as Public for team-wide access
Periodic Report Automation
Periodic Report Automation
- Compose monthly report panels in the dashboard
- Integrate HTML export with scheduled tasks
- Auto-generate HTML reports and email every Monday morning
Model Usage Analysis
Model Usage Analysis
- Generate charts via “Compare per-model token usage and cost”
- Grouped Bar for per-model · per-period comparison
- Analyze across last 7 days/30 days/all via period filter
Related Pages
Database (DbSphere)
Configure databases that BI Dashboard connects to
Schedules
Scheduler for automating dashboard exports
Usage
Token usage and cost statistics dashboard