Dynamic AI Visualization Engine

Open Source by GetCognition 💚

A visualization engine that allows AI/LLM systems to programmatically request and generate custom visualizations to explain insights.

✨ Features

• 🎨 9 Chart Types: Bar, Line, Scatter, Pie, Gauge, Funnel, Heatmap, Radial, Timeline
• 🤖 AI-First Design: Natural language interface for AI to request visualizations
• 📚 Vega-Lite Output: Standards-based, renders anywhere
• 🎯 Customizable: Easy to override colors and styling

📦 Installation

pip install dynamic-viz

For full features (recommended):

pip install dynamic-viz[altair]

🚀 Quick Start

from dynamic_viz import AIVizAssistant

assistant = AIVizAssistant()

# AI requests a comparison chart
spec = assistant.visualize_comparison(
    title="Competitor Threat Levels",
    categories=["Crayon", "Klue", "Kompyte"],
    values=[8, 7, 5],
    insight="Crayon is the primary threat"
)

# Render with vega-embed in browser or save as JSON
print(spec)

📊 Available Methods

Comparison (Bar Chart)

spec = assistant.visualize_comparison(
    title="Market Share",
    categories=["Microsoft", "Google", "Amazon"],
    values=[35, 30, 25],
    insight="Microsoft leads the market"
)

Trend (Line Chart)

spec = assistant.visualize_trend(
    title="Revenue Growth",
    dates=["2024-Q1", "2024-Q2", "2024-Q3", "2024-Q4"],
    values=[100, 120, 135, 150],
    insight="Consistent 15% quarterly growth"
)

Distribution (Pie/Donut)

spec = assistant.visualize_distribution(
    title="Market Segments",
    categories=["Enterprise", "SMB", "Consumer"],
    values=[50, 35, 15]
)

Single Metric (Gauge)

spec = assistant.visualize_metric(
    title="Customer Satisfaction",
    value=85,
    max_value=100,
    insight="Above target of 80"
)

Funnel

spec = assistant.visualize_funnel(
    title="Sales Pipeline",
    stages=["Leads", "Qualified", "Proposals", "Closed"],
    values=[1000, 500, 200, 50]
)

Custom (Any Vega-Lite Spec)

spec = assistant.visualize_custom({
    "title": "Custom Chart",
    "chart_type": "scatter",
    "data": [{"x": 1, "y": 2}, {"x": 3, "y": 4}],
    "x_field": "x",
    "y_field": "y"
})

🎨 Customization

from dynamic_viz import BRAND

# Override colors
BRAND["primary"] = "#00F5FF"
BRAND["secondary"] = "#E000FF"
BRAND["background"] = "#111827"
BRAND["text"] = "#e2e8f0"

🔧 API Reference

AIVizAssistant

Method	Description
visualize_comparison()	Bar chart for category comparison
visualize_trend()	Line chart for time series
visualize_distribution()	Pie/donut for proportions
visualize_metric()	Gauge for single value
visualize_funnel()	Funnel for process stages
visualize_custom()	Any chart from dict spec
get_chart_history()	Get all generated charts

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

📄 License

MIT License - see LICENSE for details.

💚 Credits

Open sourced with love by GetCognition

Built With

• Altair - Declarative statistical visualization library for Python
• Vega-Lite - Grammar of interactive graphics
• Vega - Visualization grammar

Special thanks to the Altair team for their amazing work on declarative visualization! 🙏