Skip to content

Consumption Layer

The consumption layer (gold) is where curated data becomes business-ready. Use it after finalizing your curated mappings, when you want to shape data into entities that analysts and downstream applications consume — for example customer_360, order_summary, or product_catalog. Nexa’s AI generates initial consumption entities from your curated layer, and you refine them through a table view or a visual canvas.

Navigate to Canvas in the sidebar (or continue from the Curated Zone). You’ll land on the consumption landing page, which offers two views.

A paginated card list showing entity pairs:

Column What It Shows
Curated Entity Source curated table name with a count badge showing the number of source tables
Consumption Entity The generated business entity name with a status badge and AI-generated description

Status badges indicate where each entity stands:

Badge Meaning
Draft Entity created but not yet finalized
AI-generated Entity was auto-generated and hasn’t been manually reviewed
Success Entity has been reviewed and artifacts generated

Each row has an AI Analyze button that triggers deeper AI analysis of the entity mapping.

Toggle to the visual canvas for a diagram-based experience. The screen splits into two resizable panels:

  • Header panel — a collapsible header with workspace tabs for SQL editing, sample records, ETL transformations, and NLP-to-code generation
  • Canvas panel — the interactive consumption canvas showing entities as connected nodes

Expand any entity row (table view) or select a node (canvas view) to open the mapping editor. Each column shows:

Element Description
Column Name The consumption entity column (read-only display with the full qualified name)
Data Type Badge showing type and nullable status
Description button Opens a side panel for viewing or editing the column description

New columns that need attention are highlighted:

  • Complete — the column has a name, type, and description (shown with a green border).
  • Incomplete — the column needs review (shown with a yellow border and a warning icon).

The visual canvas shows your data lineage across layers:

  • Curated entities appear as purple nodes.
  • Consumption entities appear as green nodes.
  • Field-level mappings are shown as dashed lines connecting individual columns between entities.
Action How
Drag nodes Click and drag to rearrange the layout
Collapse/Expand Toggle the arrow to show or hide a node’s column list
Create field mapping Drag from a curated column handle to a consumption column handle
Edit column name Click to rename consumption columns inline
Delete column Use the remove button on individual consumption columns

When you rename a consumption column, the system automatically suggests matching curated columns based on name similarity. AI-generated mappings are marked with an autoGenerated flag and include a reason explaining why the match was suggested.

Mapping types include:

  • Direct — one-to-one column mapping
  • Expression — transformation-based mapping
  • Aggregate — aggregation function applied

The collapsible header in canvas view provides additional tools:

Tab Purpose
Workspace SQL editor for writing custom transformations
Sample Records Generate and preview sample data from tables
ETL Transformations View generated transformation code
NLP-to-Code Describe a transformation in plain English and get generated code

From the consumption landing page header:

Action What It Does
Generate Artifacts Creates consumption-layer schema files, transformation code, and pipeline configs. Progress is polled and displayed in a modal.
Check Diff Shows schema differences between branches for review
View Classifications Opens the classification and tagging view
My Changes Shows your pending modifications

Search is available with a debounced input — type to filter entities by name.

After finalizing your consumption entities:

  • Select Generate Artifacts to produce pipeline code.
  • Resolve any conflicts detected during mapping.
  • Visualize the full data model on the Canvas.
  • Deploy through Pipelines.