Wiki / AI Features & AutoMode

AI Features & AutoMode

How to use the AI Chat, prompt effectively, and leverage AutoMode for autonomous dashboard building.

AI Chat Panel

The AI Chat panel is your primary interface for interacting with the AI assistant. Open it by clicking the chat icon in the toolbar or pressing Ctrl+Shift+C.

What you can do:

Create new containers: "Add a bar chart showing sales by region"
Modify existing containers: "Change the color scheme to blues"
Fix errors: "The chart isn't rendering, please fix it"
Ask questions: "What chart type would best show this data?"
Complex requests: "Create a dashboard with 4 charts comparing Q1 vs Q2 performance"

Prompt Tips

Be specific about data

Instead of "make a chart," say "create a bar chart with categories A, B, C having values 10, 25, 15."

Reference container types

Mention the library you want: "Create a Plotly 3D scatter plot" or "Use Mermaid to draw a flowchart."

Describe the outcome

"I want a dashboard showing our sales pipeline with a funnel chart, monthly trends line chart, and a summary table."

Iterate naturally

After the AI creates something, refine: "Make the bars wider," "Add a legend," "Change the title to Revenue Overview."

Tip: The AI sees your entire dashboard context. You can say "modify the second chart" or "update the Vega-Lite container" and it knows which one you mean.

AutoMode

AutoMode enables the AI to autonomously build and refine your dashboard without requiring approval for each step.

How it works:

Enable AutoMode from the chat panel toolbar (toggle switch or Ctrl+Shift+A)
Set the maximum number of iterations (default: 5)
Describe what you want: "Build a complete sales analytics dashboard"
The AI will iteratively create containers, adjust layouts, fix rendering issues, and refine the result
You can stop AutoMode at any time by clicking "Stop" or pressing Escape

AutoMode settings:

Max iterations — How many rounds the AI can take (1-20)
Auto-apply — When enabled, suggestions are applied immediately without confirmation
Error recovery — AI automatically detects and fixes rendering errors

Tip: Start with a clear, high-level description. AutoMode works best when given a goal like "Create a financial dashboard with revenue, expenses, and profit margin charts using this CSV data."

How Suggestions Get Applied

When the AI generates a response:

Code blocks in the AI response are detected and parsed
Each code block is matched to a container type (Vega-Lite, Plotly, Mermaid, etc.)
In manual mode, you see a preview with "Apply" / "Reject" buttons
In AutoMode, suggestions are applied automatically
Applied suggestions update the dashboard in real-time
All changes are tracked in the undo history — you can always revert

AI-Powered Features Summary

Natural language chat — Describe what you want in plain English
Context-aware — AI understands your full dashboard and its data
AutoMode — Autonomous multi-step dashboard building
Error detection — AI identifies and fixes rendering issues
Data analysis — AI can analyze imported data and suggest appropriate charts
Code generation — Generates valid specs for all 16+ container types