Pipelines

Create a Pipeline

Creating a Pipeline is a three-step dialog. When you finish, Kenseme creates the destination schema, builds a first draft of every table and column mapping, and opens the mapping editor.

In the app: Data Model › Pipelines › New Schema Pipeline

[screenshot]
The New Schema Pipeline dialog on Step 1, with one source schema selected and the naming conventions filled in.

Before you start

  • At least one source schema with tables or views. See Schemas overview.
  • Your organization must not be read-only. In a read-only organization the create button is disabled.
  • Decide which database system the destination targets. Load generation works for SQL Server, Fabric Warehouse, Databricks, and Fabric Lakehouse. See Supported targets.

Step 1: Source & Destination

  1. On the Pipelines page, click New Schema Pipeline.
  2. In Source Schema(s), pick one or more source schemas.
  3. Enter a Pipeline Name, for example ERP Bronze to Silver.
  4. Enter a Destination Schema Name, for example ERP Silver. Kenseme creates a new schema with this name.
  5. Optional: add a Description.
  6. Pick the Database System for the destination. The default is Sql Server.

  7. Set the Naming Conventions for the destination:

    Field What it does
    Table Casing PascalCase, CamelCase, or SnakeCase for destination table names.
    Column Casing The same choices for destination column names.
    Table Prefix Text added to the front of every destination table name.
    Plural Table Names Turn on to pluralize destination table names.
    ID Column Suffix Suffix for key columns. The default is Id.

  8. Click Next: Select Tables.

Kenseme stops you here if no source schema is selected, or if the Pipeline name or destination schema name is empty.

Tip: These conventions become the destination schema’s naming settings. Kenseme uses them to propose every destination table and column name, so set them before you continue.

Step 2: Select Tables & Views

The grid lists every table and view in the selected source schemas, all selected by default. The Preview column shows the destination table name your conventions produce.

  1. Clear the rows you do not want. Use Select All or Deselect All for bulk changes.
  2. Check the count below the grid, for example 12 of 14 table(s)/view(s) selected.
  3. Click Next: Documentation.

You must keep at least one table or view selected.

🚩 TODO: In the current build, a new Pipeline maps every table and view in the source schemas, not only the rows selected in Step 2. Until this is fixed, clear the include checkbox for unwanted tables in the mapping editor after you create the Pipeline.

Step 3: Documentation (optional)

This step gives the AI context for suggesting readable column names. Skip it if your source names are already clear.

Field What to put in it
Source System Documentation Data dictionaries, column descriptions, entity contracts. More context gives better suggestions.
Additional Instructions Naming rules, for example Use Customer instead of Client or Date columns should end with Date not Dt.
Run AI Rename Tick it to have the AI suggest destination column names after the Pipeline is created.

Then click Create Pipeline.

What happens next

  • Kenseme creates the destination schema and applies your naming conventions to it.
  • It creates a table mapping for each source table and view, and a column mapping for each column. Every table starts with the Truncate & Reload load strategy.
  • Each source view gets an extra key column, named after the destination table plus Id (for example CustomerSummaryId), because the view becomes a real table in the destination.
  • The mapping editor opens.

If you ticked Run AI Rename, the rename runs in the background after the editor opens. Suggestions appear in the Table Mappings tab when it finishes. Reload the page to see them, or follow the job in Where to watch background jobs.

Note: If your organization has used its monthly AI allowance, the Pipeline is still created but the AI rename is skipped. You can run it later from the mapping editor’s Documentation tab. See AI token allowance and usage.