Skip to main content

Node Graph

Visual representation of your analytical pipeline—see how data flows from sources to insights.

What is the Node Graph?​

The node graph is the visual map of your analysis. Every Source, View, AI Table, and Input Object appears as a colored node, with lines showing how data flows between them. It's like a flowchart of your entire analytical process, automatically generated as you work.

Node Graph Overview

Visual representation of data flowing through transformations

Why the Node Graph Matters​

Understand your pipeline: See the complete path from raw data to final insights.

Debug visually: Identify where issues occur by tracing through nodes.

Verify logic: Check that data flows make sense before trusting results.

Learn SQL patterns: See how the AI breaks complex queries into steps.

Collaborate better: Show teammates how analysis was built.

Node Colors and Meanings​

Each node type has a distinct color:

Blue Nodes - Sources

  • Your original data (CSV, database tables)
  • Starting point for all analysis

Colored Nodes - Views (by type)

  • Filter, Calculate, Aggregate, Combine, etc.
  • Each operation type has its own color
  • See Views for complete type list

Purple Nodes - AI Tables

  • LLM-enhanced data with intelligent columns
  • Manually refreshed

Yellow Nodes - Input Objects

  • Variables and scenario parameters
  • User-configurable values

Node Color Legend

Each node type has a unique color for easy identification

Understanding Data Flow​

Reading the Graph​

Left to right: Data generally flows from Sources (left) to final results (right)

Connections: Lines show dependencies—arrows point to nodes that depend on earlier ones

Branches: When one View splits into multiple downstream Views, you'll see branching paths

Convergence: When multiple Views feed into one (like joins), lines converge

Data Flow Example

Example: Orders and Customers joining, then aggregating to monthly revenue

Node Information at a Glance​

Each node shows:

  • Name: What the View or Source is called
  • Icon: Classification icon (for Views)
  • Badge: Primary operation type
  • Summary: 4-word description of logic
  • Dimensions: Row × column count

Click any node to see full details in the bottom panel.

Interacting with the Graph​

Clicking Nodes​

Single click: Select node and show data preview below Double click: Expand node to see detailed explanation Right-click: Open context menu with actions

Node Context Menu

Right-click menu for node operations

Pan: Click and drag the background to move around Zoom: Scroll to zoom in/out Fit view: Button to reset and show all nodes Auto-layout: Nodes arrange automatically as you create them

Highlighting Dependencies​

When you select a node:

  • Upstream dependencies (parents) highlight
  • Downstream dependents (children) highlight
  • See the full data lineage at a glance

Dependency Highlighting

Selected node highlights its upstream and downstream connections

Common Use Cases​

Tracing data lineage: Click final result and see what Sources it came from

Debugging transformations: Find where unexpected results originated

Understanding complexity: See how many steps the AI used for complex analysis

Identifying bottlenecks: Large nodes with many dependents might need optimization

Learning by example: Study how the AI breaks down requests into Views

Explaining to stakeholders: Show the analytical process visually

Node Details Panel​

Click any node to see:

Data Tab: Preview of actual rows and columns Profile Tab: Metadata and statistics Code Tab: SQL query (for Views) Explanation: What this node does and why

Node Details

Bottom panel showing detailed node information

Graph Organization Features​

The left sidebar complements the graph:

  • Sources section: Lists all blue nodes
  • Views section: Lists all transformation nodes (with count)
  • AI Tables: Purple nodes separately listed
  • Inputs: Yellow variable nodes
  • Visualizations: Charts attached to nodes

Click any sidebar item to locate and select its node in the graph.

Search Functionality​

Use the search box to quickly find:

  • Views by name
  • Sources by name
  • Specific transformations

The graph highlights matching nodes.

Search Interface

Search to locate specific nodes quickly

Grouping Nodes​

You can group related nodes for organization:

  • Select multiple nodes
  • Right-click and choose "Group"
  • Name the group for easy reference

Grouped nodes can be collapsed to simplify complex graphs.

Tips & Best Practices​

Use the graph to verify joins: Check that data combinations make sense before analyzing results.

Trace unexpected results backwards: Click the problematic node, then check each upstream node to find where things went wrong.

Clean up unused Views: Delete nodes you no longer need to keep the graph readable.

Name Views meaningfully: Good names make the graph self-documenting.

Check node dimensions: Unexpected row counts often indicate issues (like missing joins or incorrect filters).

Expand nodes for context: Double-click Views to see detailed explanations.

Use groups for complex analyses: Organize related transformations to reduce visual clutter.

Understanding the Reactive System​

The graph isn't just visual—it's functional:

  • Change a Source → all downstream Views recalculate automatically
  • Edit a View → everything depending on it updates
  • The graph shows this reactivity in real-time

See Reactive System for details on automatic updates.

Reactive Updates

When one node changes, downstream nodes automatically update