Create and configure a Dataset

A Dataset is a semantic layer of data that enhances self-service capabilities in the Board platform. It allows users to interact with data without needing to know in detail all the data structures of a Data Model.

This articles covers:

How to create a Dataset

From the Dataset Catalog, Developers can create and manage Datasets.

To create a Dataset:

Access the Dataset Catalog.
Click on the plus icon situated on top of the Dataset Catalog.
Define all the Dataset properties on the "New Dataset" panel.
Dataset Name. Name of the Dataset to be displayed in all other functionalities where Datasets is involved.
Dataset Description. Developer-defined explanation of the Dataset.
Datasets and Board AI Agents
For the best experience with Board AI Agents, include functional purpose, measures included, key dimensions, and intended use cases for the Agent to understand when to use this Dataset.
- Remember you’re writing for finance users, not for developers
- Include: scope, granularity, time periods, functional purpose of use
- Avoid: acronyms without any context, table names, cryptic cube names, ETL notes, and other unexplicit content
Example:
Dataset Name: P&L Actuals and Budget (Monthly)
Dataset Description: Income statement measures by Legal Entity, Cost Center, Account and Month. Includes Actual, Budget and Forecast versions. It provides a structured view of financial performance over time (revenues, costs, margins, operating profit). Essential for variance and driver analysis across periods, scenarios, and business dimensions.
Datasets and Board AI Agents
To ensure consistent interpretation and correct dataset selection, relevant business context should be clearly defined across both:
- Dataset metadata (Dataset Name & Description)
- Agent configuration (e.g., Knowledge Base, uploaded documents, fine-tuning inputs)
The Dataset Description should provide a concise, high-level overview of the data exposed.
Include (when relevant):
- Source system or data domain (ERP vs CRM vs custom)
- High-level scope and purpose of the dataset
- Key inclusions/exclusions (e.g., IC eliminated, adjustments, entity scope)
- Currency context (e.g., Local vs Group)
- Data freshness (e.g., daily close, D+1, intraday)
Detailed logic such as mappings, normalization rules, or complex assumptions should not be fully embedded in the dataset description, but instead provided via:
- Advance fine tuning area (e.g., solution manual, contextual files)
Dataset for AI. Enable if this Dataset will be used by Board AI Agents
Data Configuration. Data Block configuration. Learn more about the Quick Layout.
By design, you can copy from Dataset to Screen but not the other way around
Data Configuration for Board AI Agents
- Use clear intuitive Block Headers (mandatory): The Agent uses Block Headers to understand the semantic meaning of each block (e.g., Actual, Budget, Variance, Drivers).
If headers are unclear or inconsistent, the Agent may misinterpret the data or produce incorrect explanations.
- Keep Block structures aligned at the right granularity level: Align Blocks so the Agent can compare and drill consistently across the Dataset.
The key requirement is not identical cube design. The key requirement is that the Agent receives data at a comparable and interpretable granularity for the use case.
Avoid implying that all Blocks must use the exact same Entities or hierarchies. What matters is that the Dataset structure supports consistent comparison, drill paths and interpretations for the target use case of the user.
- Cube versioning: You can leverage cube versioning to speed up performances. Using aggregated cube versions: reduces query execution time and improves responsiveness for top-level questions.
The Axes tab
- Choose one main Entity in “By Column”: Keep the view readable and consistent for analysis and commentary.
- Limit “By Row” Entities (max 2–3): Too many rows increases Layout size and risks increasing the number of combinations for the agent to evaluate.
- Use “Show all” only when needed: Disable Show all by default and enable it only for specific use cases (e.g., rule-based calculations).
- Quick Layout behavior: The Agent can ONLY understand and manipulate what is explicitly defined in the Quick Layout — including axes, axis settings, and drill paths. The Agent cannot infer structures, drill paths, or selectable dimensions that are not defined there. If something is not exposed in the Quick Layout, the Agent cannot reliably use it.
The Select tab
- Use Selections to define the Dataset perimeter: Apply required selections to filter the Dataset (scope guardrails). The Agent can read these Selections but not override them preventing it from mixing scopes and ensure consistent analysis and commentary.
- Use Selections to enforce Entity scope (e.g., Group only, Region, specific business units), scenario/version (Actual, Budget, Forecast), Currency / reporting view (Local vs Group), Inclusions/exclusions (e.g., ex-IC, adjusted view).
Rules and governed Entities
If an Entity has rules applied, mark it as:
- Mandatory (always present in Agent queries)
- Most nested (rules are applied at the correct level of detail)

Other considerations
- The Quick Layout defines the hierarchies and drill paths the Agent can use
- The Agent can only work with the entities exposed in the Quick Layout of the linked datasets
- Include only what is needed for your use cases to keep answers stable and relevant
- The Agent only receives the Entities explicitly configured in the Dataset Quick Layout
- Entities must be clearly set as: By Row, By Column, Mandatory, Nested
- These settings control how the Agent can pivot and reconfigure the Layout when the user requests a different view
- Do not add Entities that are already enforced as Dataset Selection filters to the Quick Layout (they are fixed scope, not best used as pivot axes).
Click "CREATE".

Interact with Datasets in Board

Datasets are not visible in Capsules or Screen Objects. Users interact with Datasets through other features in Board.

Board AI Agents

The Datasets feature will play an important role in allowing the Developer to prepare a set of queries that represent the best data to be used by Board AI Agents.

Datasets define data context for Board AI Agents.

Developers create Datasets optimized for AI responses.

Board M365 Excel Add-in

The new Excel Add-in defaults to using Datasets for data extraction from Board, leveraging the same backend technology utilized by the Flex Grid Object. The user will do all the data manipulation inside Excel. Users using the Excel M365 Add-in simply need to pick the Dataset that they wish to use and go directly to the pivoting stage to create custom views – there is no need to use the Layout editor.

Board Foresight

The purpose of a Dataset in Board Foresight is to enable the sharing of specific data with Foresight to enable the user to create a forecast which will then be returned to Board, automatically stored in a Cube and made available for the Developer to add to Screens.

Enables secure, structured data sharing with Foresight.
Forecasts are generated in Foresight, returned to Board, and stored in Cubes.
Data Models are abstracted from the API interface.

Datasets known limitations
Any key change like Entity renaming will require a recreation of the Dataset - Already created datasets will not adapt to the key data structure changes made after creation
Data Entry, Detail By & Total By are not supported in the Layout settings of a Dataset
Nexel, Dynamic Options, Sparklines, Column Appearance are not visible in Datasets