Training Ask Me

Ask Me generates responses based on pre-trained models and the available data within the platform. While this enables users to ask questions and receive insights quickly, the system currently does not allow users to directly provide feedback or train the model based on their specific business needs.

To improve the accuracy and relevance of responses over time, Ask Me will introduce a training and feedback capability that allows users to guide the AI by correcting responses and providing feedback. This will help the system learn from user training and deliver more personalized and accurate results.

Benefits

Once implemented, the training capability will enable users to:

• Improve the accuracy of Ask Me responses by providing direct feedback and corrections.
• Adapt the AI to their specific business context, ensuring insights are more relevant to their data and use cases.
• Minimize recurring errors in generated queries, visualizations, or insights.
• Continuously improve Ask Me’s performance by enabling the system to learn and adapt from user-provided training.

To train query:

Step 1: Navigate to Home > Ask Me, then select “Settings & Help.”

Step 2: Click the “Train Queries” option.

Step 3: Select “Add Training Queries.”

Step 4: Choose the schema and enter the query you want to train.

Step 5: You can edit the SQL query.

The SQL Query Editor allows users to view, edit, or modify the SQL query used to generate the chart or analysis. This feature helps users refine queries to improve the accuracy and relevance of the output.

The editor provides the following components:

• Fields Panel: Displays the available tables and fields from the selected dataset or schema. Users can browse data sources such as Customer and Orders and easily reference them while writing queries.
• Functions Tab: Provides commonly used SQL functions that can be applied within the query.
• Write Expression Area: This is the main workspace where users can write or modify the SQL query. For example, users can calculate totals, apply filters, group data, or sort results.

After updating the query, users can click Validate or Preview to check the output and ensure the query returns the expected results.

Once input is valid, click “Save & Next.”

Step 6: This page also provides an option to edit the SQL query. By clicking Edit, you will be redirected to the previous editing page where you can modify the query.

Chart Type:
You can choose the type of visualization used to display the result. Available chart options include:

• Grid
• Column
• Spline
• Area
• Step Area
• Area Spline
• Line
• Step Line
• Bar
• Lollipop

Changing the chart type allows users to select the most suitable visualization for the data.

Preview:
The Preview panel displays the SQL query output and the selected chart type. This helps users verify that the query and visualization produce the expected result before training the model.

Share Option:
The Share option allows users with whom the schema is shared to view and use the training. However, only the schema owner can edit or add new training; shared users have view-only access.

Train Module:
Once the query, chart type, and preview are verified, users can click Train Module to save the training. This helps Ask Me generate more accurate responses when similar questions are asked in the future.

Step 7: You will return to the Training page, where all the trained queries are listed. From here, you can edit or delete any trained query as needed. Once the training is complete, you can start by asking the queries you have created in Ask Me to generate the corresponding insights.