Understanding the AI-Generated UI — Metrics, Tables, Charts & Dropdowns
Learn how Unsheet's AI analyses your spreadsheet and generates the perfect combination of metric cards, interactive tables, charts, and filter dropdowns.
The Four Component Types
Unsheet's AI generates apps using four component types, each optimised for a specific kind of data:
📊 Metric Cards
Large, prominent numbers that display key totals or KPIs. The AI identifies cells with =SUM() or similar aggregation formulas and creates metrics for them. Each metric card references a specific cell (e.g., D15) and updates in real-time as the underlying data changes.
📋 Interactive Tables
Your core data in a clean, scannable table. Number and currency columns are editable — change a value and watch formulas cascade. String columns are read-only, and enum-like columns get styled badges. Tables display up to 50 rows.
📈 Charts
Bar or line charts that visualise numeric columns. The AI picks the best column for labels (x-axis) and one or more numeric columns for values (y-axis). Charts support multiple series and update live when data changes. Up to 20 data points are displayed.
🔽 Dropdowns
Filter controls for categorical columns. When you select a value, it writes to a specific cell in the spreadsheet engine, potentially triggering formula recalculation. Perfect for scenario modelling and "what-if" analysis.
How the AI Analyses Your Data
When you upload a file, Gemini receives:
- Column names and data types — "Item" (string), "Quantity" (int64), "Price" (float64), "Total" (float64)
- The first 5 rows of sample data
- Total row count
From this context, the AI determines:
- Which columns should be table columns (and their display type: string, number, currency, enum)
- Where summary/total rows exist (for metric cards)
- Which numeric columns make sense as chart data
- Whether any columns have few distinct values (for dropdowns)
The result is a UISchema — a JSON structure that tells the frontend exactly what to render and how.
Auto-Generated Theme
The AI also generates a CSS gradient that matches the "vibe" of your data. A financial spreadsheet might get a deep blue gradient, while a marketing tracker might get a vibrant purple-to-pink scheme.
This gradient is applied to the top section of your app, giving each Unsheet a unique visual identity.
Found this useful?
Share it with your team or network.
Frequently Asked Questions
How does Unsheet decide what components to show?
Can I change the AI-generated layout?
Related Guides
Upload Your First Spreadsheet to Unsheet — Complete Walkthrough
Step-by-step guide to uploading an Excel or CSV file and watching Unsheet's AI transform it into a live web application with charts and formulas.
Edit Mode — Customise Your Unsheet App After Generation
Use Edit Mode to change titles, descriptions, and add metric components to your AI-generated Unsheet app. Full customisation walkthrough.
Live Formula Recalculation — How Unsheet Keeps Your Data Alive
Unsheet uses HyperFormula to evaluate 400+ Excel functions in real-time. Edit any value and watch totals, charts, and metrics update instantly.
Ready to kill your spreadsheets?
Upload your first file and see the magic in under 30 seconds.
Get Started Free →