Skip to content
  • There are no suggestions because the search field is empty.

Understand Genie: The smart biGENIUS-X Assistant

Genie is an AI-powered chat system for biGENIUS-X that provides assistance for data modeling tasks or general questions about the application.

It combines multiple expert systems with intent detection to deliver contextually appropriate responses for data solution design and biGENIUS-X modeling questions. 

Overview

Genie includes the following capabilities:

  • Knowledge Assistance: Answers questions about existing data modeling concepts, biGENIUS-X features, and best practices. Uses retrieval-augmented generation (RAG) with vector search to provide accurate, documentation-based responses for the application
  • Model Proposals: Creates detailed data model designs based on user requirements. Generates structured proposals with model object specifications, including model objects, terms, relationships, wizard recommendations, and implementation guidance tailored to the user's chosen modeling methodology.
It routes the queries to different expert paths:
  • Ask the doc: all the biGENIUS-X Knowledge Base content
  • Data Vault expert: Data Vault modeling expertise around the Web
  • Dimensional expert: Dimensional modeling expertise around the Web
  • Data Vault Data Model Proposal: Data Model expertise for Data Vault modeling
  • Dimensional Data Model Proposal: Data Model expertise for Dimensional modeling

Use Genie

If you click on the Genie icon located on the right-hand side of the navigation bar, it will display the list of notifications for the connected user.

The Genie side panel opens:

It is composed of 3 parts:

  1. Genie header
  2. Genie content
  3. Genie chat input

Genie header

In the Genie header, you can:

  • Switch between the Knowledge assistance and the Model Proposal:
  • Expand the Genie side panel by clicking on the Expand button:
  • Collapse the Genie side panel by clicking on the Collapse button:
  • Start a new chat session by clicking on the New Chat button:

Genie content

Here is displayed the answer to a question depending on the used capability.

By clicking on an example question or filling a prompt in the chat input, Genie starts working on it:

Then, it displays the answer:

At the bottom of the answer, you can see which path was used by Genie for your question:

You can change it to have an answer based on a different path:

The proposed paths are related to your Project Generator Configuration. In this example, we are using a Generator with Data Vault modeling. The Dimensional modeling path is not proposed as it is not relevant.

Knowledge assistance

When opening Genie or starting a new chat for Knowledge assistance, some example questions are here:

Depending on the path used by Genie, the answer is different:

  • biGENIUS Knowledge base: 
    • The answer is an extract of the Knowledge Base
    • Some links allow you to open the relevant articles
  • Data Vault / Dimensional expert:
    • The answer is an extract of the Web, not related explicitly to biGENIUS-X

Example:

  • Prompt:
    • Which layers must include a good data vault architecture?
  • biGENIUS Knowledge base answer:
  • Data Vault expert answer:
Model proposal

When opening Genie or starting a new chat for Model Proposal, some example questions are here:

The path used by Genie is: Data Vault Data Model Proposal.

Example:

  • Prompt:
    • If I have a source containing customer data, which object must I model in the raw vault area? 
    • Answer:

The modeling proposal is always related to your Project Generator Configuration. In this example, we are using a Generator with Data Vault modeling.

So, the answer includes Hub and Satellite, and not Fact and Dimension.

Some questions are not relevant for an expert or another. 

In this case, Genie will display the following message: "This question is not related to my area of expertise as a XXX expert."

Genie chat input

In the Genie chat input field:

You can enter your question.

In a Project, the Project context is known by Genie. You can include a Model Object in your question by using the key character @:

Then, you can select one or the other existing Model Object in your Project.

Some advice to have a good prompt:

  • Be Clear and Specific: Define exactly what you want from the AI. Avoid vague requests.

  • Use Natural Language: Write as if speaking to a person for better understanding.

  • Provide Context: Include relevant background or details to guide the response.

  • Keep It Concise: Provide sufficient details but avoid unnecessary complexity.

  • Define the Format: Specify the output type if needed.

  • Iterate and Refine: Adjust prompts based on results to improve answers.

Demonstration video