Data Sources

Use this page to understand how Data Sources are used, configured, and governed in AI Studio.

Overview

Data Sources define the governed datasets available to AI Studio.

A Data Source connects specialists and AI workflows to approved Yarken data. It controls what data can be queried, which columns are available, and whether user-level authorization is applied.

The Data Sources page describes this area as the place where warehouse tables and query tools are available to the agent.


User flow

Users do not directly select Data Sources when asking questions.

Data Sources are used behind the scenes when Ask Yarken, a Specialist, Playbook, or Automation needs data to answer a question or run a workflow.

User action

What happens

User asks a question in Ask Yarken

The orchestrator determines whether data is required.

A Specialist is selected

The Specialist can use only its assigned Data Sources.

Data is queried

User authorization is applied where enabled.

Results are returned

The Specialist or workflow uses the data to generate the response.


How Data Sources work

8c558f45-84e9-4eec-8c8f-898e062253a6

Where to find Data Sources

To open Data Sources:

  1. Go to Admin.

  2. Select AI Studio.

  3. Under Configure, select Data Sources.

The page displays data source cards in the main panel.

Screenshot 2026-07-09 at 8.53.10 PM.png

Default data sources

For new users, built-in data sources are loaded by default.

The current default data sources include:

Data source

Status

Purpose

Spend report

Active

Query spend data by vendor, cost centre, category, period, and organizational hierarchy. Supports dynamic dimension and measure selection with fiscal year awareness.

Cloud spend report

Active

Query cloud spend data using FOCUS metrics. Supports provider, service, resource, tag-based, billing, and cost breakdown dimensions.

Contracts report

Active

Query vendor contracts, including contract values, PO details, invoice amounts, and expiry dates.

Spend line items report

Active

Query GL spend line items for transaction-level detail, including journal lines, invoice numbers, line descriptions, and similar details. Use only when transaction detail is required.

Chargeback Analysis

Active

Analyze IT service chargeback data to compare allocated costs against actual consumption by business unit.

BU Showback Report

Active

Shows business unit cloud application costs and service allocations for showback analysis.

Asset Analysis Report

Active

Analyze IT assets, including software licenses, hardware, and related costs by asset type, code, and name.

License Usage

Active

Analyze software license usage and related license details.

The default list may change before release.


What each data source card shows

Each data source card shows:

Field

Description

Data source name

User-facing name of the data source.

Status

Whether the data source is active.

Description

Summary of what the source is used for.

Actions menu

Options such as Edit, Deactivate, and Delete where available.


How Data Sources appear in instructions

When a Data Source is assigned to a specialist or used in an AI workflow, it can be referenced in instructions using @.

For example:

Use @Spend_report when the user asks for aggregate spend by vendor, cost centre, category, period, or organisational hierarchy.

Use @Spend_line_items report only when the user explicitly asks for transaction-level or line-item detail.

Use @Contracts_report when the user asks about contract values, PO details, invoice amounts, or expiry dates.

The @ mention picker may show Data Sources alongside system Tools. These are separate groups.

A Data Source controls what data can be queried.

A Tool controls what action the agent can perform.


Admin configuration flow

8d1986a6-c4e8-4ec9-8d34-8a5b9c70342f

Add a data source

To add a data source:

  1. Go to Admin > AI Studio > Data Sources.

  2. Select + Add Source.

  3. Enter the identity details.

  4. Configure the data source columns and options.

  5. Confirm authorization behavior.

  6. Select Save.

Detailed setup steps should be confirmed after the Add Source workflow is reviewed.


Import a data source

Use Import to upload a data source configuration from a local file.

To import a data source:

  1. Go to Admin > AI Studio > Data Sources.

  2. Select Import.

  3. Drag and drop the file into the upload window, or select Browse.

  4. Upload the file.

Screenshot 2026-07-09 at 9.34.59 PM.png

Supported file types:

  • .xlsx

  • .csv

After the file is uploaded, Yarken shows a data preview. Review the imported fields before continuing.

Screenshot 2026-07-09 at 9.36.04 PM.png

If required fields are detected successfully, Yarken automatically maps them.

You can then:

  • Select Finish to complete the import.

  • Select Edit Mapping to review or adjust field mapping.

  • Select Save field mapping for automation if the same mapping should be reused later.

Screenshot 2026-07-09 at 9.36.30 PM.png

Export a data source

Use Export to download a data source configuration as a CSV file.

To export a data source:

  1. Go to Admin > AI Studio > Data Sources.

  2. Open the action menu on the data source card.

  3. Select Export.

Screenshot 2026-07-09 at 9.54.03 PM.png

The export starts immediately. Yarken downloads the data source information as a .csv file.


When to use import and export

Use import and export when you need to:

  • Move a data source configuration between environments.

  • Create a data source from an approved configuration file.

  • Review data source metadata outside Yarken.

  • Validate field names, categories, descriptions, authorization settings, and default status.

  • Keep a record of data source setup before making changes.


View or edit a data source

To view or edit a data source:

  1. Go to Admin > AI Studio > Data Sources.

  2. Open the data source card.

  3. Review the identity, data source columns, and options.

Screenshot 2026-07-09 at 9.00.00 PM.png
Built-in data sources cannot be modified or deleted
Screenshot 2026-07-09 at 9.00.28 PM.png
  1. Update editable fields if required.

Screenshot 2026-07-09 at 9.01.59 PM.png
Screenshot 2026-07-09 at 9.03.25 PM.png
  1. Select Save.

Screenshot 2026-07-09 at 9.04.48 PM.png

Some data sources are built in and managed by the system. These may appear as read-only.


Data source identity

The Identity section defines how the data source appears and how the agent understands its domain.

Field

Required

Description

Display Name

Yes

The user-facing name. Example: Spend report.

Description

Yes

Explains what the data source provides and when it should be used.

Category

Yes

A short label that groups the source with related tools. Common values shown include spend, cloud, contracts, and hr.

Use a clear category so agents can understand the data domain.


Data source columns

The Data Source section lists the columns available to the agent.

Each column includes:

Field

Description

Name

Internal column or field name.

Kind

Whether the field is a dimension or measure.

Description

Business meaning of the field.

Examples shown in the screenshots include:

Column

Kind

Description

date

Dimension

Transaction date. Prefer year, quarter, or month dimensions for fiscal-period questions unless using an explicit date range.

year

Dimension

Financial year label stored in Yarken.

quarter

Dimension

Fiscal quarter within the financial year.

month

Dimension

Month within the financial year.

costCentre

Dimension

Cost center name.

costCentreCode

Dimension

Cost center code.

account

Dimension

Account name.

accountCode

Dimension

Account code.

entity

Dimension

Entity name.

entityCode

Dimension

Entity code.

spend

Measure

Total spend amount.

budget

Measure

Total budget amount.

forecast

Measure

Total forecast amount.

The number of available columns depends on the data source. Examples shown include 63 columns for Spend report and 834 columns for Asset Analysis Report.

Screenshot 2026-07-09 at 9.07.33 PM.png



Row-level authorization

The Apply user authorisation option controls whether results are limited to data the current user is authorized to see.

When enabled, row-level security limits results to authorized data.

Keep this enabled unless the data source is specifically intended to expose organization-wide aggregates to all permitted users.


Activate or deactivate a data source

Active data sources are available to the agent.

To deactivate a data source:

  1. Go to Admin > AI Studio > Data Sources.

  2. Open the action menu on the data source card.

  3. Select Deactivate.

Use deactivation when the data source should remain configured but should not be used.


Delete a data source

To delete a data source:

  1. Open the data source.

  2. Scroll to Delete data source.

  3. Select Delete.

Screenshot 2026-07-09 at 9.10.55 PM.png

Deleting a data source permanently removes the data source and its configuration.

Do not delete built-in or shared data sources unless you are certain they are not used by any specialist, skill, playbook, or automation.


Configuration guidance

Before making a data source available, confirm that:

  1. The source is approved for AI use.

  2. The description clearly explains when to use it.

  3. The category is accurate.

  4. Column names and descriptions are clear.

  5. Dimension and measure types are correct.

  6. Authorization behavior is appropriate.

  7. Assigned specialists actually need the source.


Next step

Skills