Revision Date: 05 December 2022
This documentation has been created for software version 2022.4.0.0
It is also valid for subsequent software versions as long as no new document version is shipped with the product.
Suite 105, 220 Davidson Ave, Somerset, NJ, USA 08873
For more information, visit
Copyright © 2022 whiz.ai All Rights Reserved.
Trademarks owned by whiz.ai
No Warranties and Limitation of Liability
Every effort has been made to ensure that this document is an accurate representation of the steps to deploy WHIZ.AI platform. However, the development of the software is a continuous process. So, small inconsistencies may occur.
We would appreciate any feedback on this document
Send comments via email
WhizAI supports the following browsers:
Google Chrome, v107 & above
Mozilla Firefox, v106 & above
Microsoft Edge, v106 & above (Windows); v106 & above (MacOS)
Safari, v15 & above (MacOS)
This document outlines the new product updates for the release version 2022.4.0.0. The following lists the new features and enhancements.
Pinboards, Cards, and Responses
Pinboards, Cards, and Responses
Actions supported on card/s in the collapse mode for all the tables and charts
Now, WhizAI supports some of the common actions on card/s in collapse mode for all the tables and charts. Particularly, you are not required to expand the card/s to perform these common actions. This helps to improve the analysis while interacting with WhizAI.
You can perform the below actions in the collapsed mode.
Reordering the columns: drag and drop the columns to manually reorder.
Pinning Column/s: lock column/s to the left side so that when you scroll either to the left or right, the pinned column/s remains at the same location.
Pagination: scroll through the response all the way to the last row irrespective of the number of records in the response.
Roll Up/Drill Down: roll up or drill down through the response details to get more information.
Expand All and Collapse All: expand or collapse the data in the nested tables.
Downloading CSV and XLS files asynchronously for all responses
Now, when you export the XLS or CSV file, you can use the WhizAI application until the file gets downloaded.
For example, when you ask ‘Show me top regions by TRx’, WhizAI shows the following response.
When you click the More Actions hamburger menu and click either Download as CSV or Download as XLS options, the following dialogue displays.
You can click OK and use the WhizAI application, once the file is downloaded you receive a notification as shown in the following figure.
Introducing a new layout for Pinboards
Now, you can view your pinboards organized in different categories, the categorization of the pinboards helps you to manage the boards and you can easily navigate to the required boards. Also, you can perform various tasks for easy access.
When you navigate to the pinboards, you will be redirected to the Pinboard Manager page. By default, the Recent section opens as shown in the following figure.
From the left-hand side of the Pinboard Manager page, you can see the various options for easy access.
Favorites: This section lists the pinboards that are starred in the Pinboard Manager section.
Recent: This section displays the pinboards that you have visited and recently created.
My Pinboards: You can create new pinboards and import the pinboard as required.
Shared: This section shows the pinboards that are shared by other board users.
All Pinboards: This section displays the list of all pinboards.
Enhancement - ‘Apply to Board’ displays the relevant list of pinboards
Now, on cards when you click Apply to Board, the list of relevant pinboards is displayed associated with the record.
For example, when you ask a response NLQ ‘Show me top customers’ and hover the cursor over this icon against Michael Johnson and click the Apply to Board button.
The list of pinboards where the customer Michael Johnson is available are displayed as shown in the following figure.
Updated filtering behavior for pinboards
Earlier, after you added and applied filters to pinboards, the corresponding records were highlighted on the cards.
Now, after you add and apply filters to pinboards, the cards show the corresponding record only.
For example, if the Region > Mid Atlantic filter is added and applied, the cards show the following result:
Insights - Option to run anomaly detection on trend
Now, you can run anomaly detection on a card for NLQs that show a trendline with weekly, monthly, quarterly, and yearly data for metrics and scope.
For example, if you ask ‘Show me TRx weekly’ then you get a response as shown in the following figure.
Click the Insights bulb icon, the Insights dialogue opens.
Enable the Anomalies button and click Apply.
When any of the data points on the trendline are outlying from the usual range, then it gets highlighted in accordance with the algorithm as shown in the following figure.
When you hover the cursor over these data points, it displays the Anomaly Details.
Insights - Creating and managing anomaly templates for various metrics
You can create and manage templates for various metrics using algorithms. When you create a template for a metric and ask a query, WhizAI understands which algorithm to run for the metric that is asked in the query to show the correct response.
To create a template, follow the steps listed below:
Click User Profiles > Insight Workbench option from explorer. By default, the Anomalies tab opens. Now fill in the details in the columns of this tab.
In the General tab, add the following details.
Metric & Anomaly Type: In this column add Data Model (Field Analytics), Metric (TRx) and Anomaly type (Four Period Moving Average).
Filters: In this column add filters to be applied on scope.
Training Dataset: In this column add Training data scope (last 104 weeks), and Granularity (Weekly).
In the Advanced tab, add the following detail.
Algorithm: Four Period Moving Average - Single point outlier based on the previous four period moving average.
Parameters: The number of standard deviations of delta away from the mean, to be considered for the anomaly. A possible value is between 1.00 and 3.00. The default is 3.00 (Add it as 3).
Click Save as Template.
After you click Save as Template, you receive a dialog that shows added details. Add the Template Name.
Click the Enable button.
Verify the details and click Save.
You receive a pop-up message that the template has been saved successfully.
The template for metric TRx is added to the Templates as shown in the following figure.
Now, when you ask WhizAI ‘Show me TRx weekly’ and click the Insights icon and enable the Anomalies from the card, you can view the anomalous behavior of the metric TRx as shown in the following figure.
Editing a template
To edit a template:
Click the Edit icon from the templates.
Edit the required parameters.
The following message pops up.
Click Yes to save the template.
Deleting a template
To delete a template:
Click the delete icon.
The following message pops up.
Insights - Viewing narratives for anomalies
Now, you can view descriptions about the anomalies on cards. These descriptions or narratives provide additional information about the anomalous data points.
For example, if you ask, ‘Show me TRx weekly’ and from the response, if you click the Narratives icon then WhizAI shows the narratives details as shown in the figure.
To view the narrative for anomalies, you have to create a Narrative Template from the Admin console.
To create a Narrative Template, follow these steps:
From the User Profile click Admin > NLP Workbench > Narrative Templates.
Click Create from the Custom Narrative Template Page.
Enter a name for the template.
Click Set Intent and select Anomaly from the Intent dropdown list.
Type the narratives in the context box,
Insights - Viewing additional information about anomalous data points
Now, you can view the following additional information about the anomalous data points:
The confidence band represents the uncertainty in an estimate of a curve.
You can see the anomaly details when you hover the cursor over any anomalous data point:
The statistic shows the statistical algorithm:
Algorithm Name: The name of the algorithm.
Coverage: It specifies what % of the total data points should be covered within the expected range prepared by the algorithm. A possible value is greater than 0 and less than 1. The default value is 0.75 (75%).
Mean of Delta: The average difference between actual and expected value.
SD of Delta: Standard deviation of the delta is the difference between the actual and expected value.
Metric: Metric Name.
Insights - Availability of the valid list of dimension filters and factors for Key Driver Analysis
Now, when you perform the key driver analysis, the only dimension filters and the factors are available for the selection that are associated with selected metrics.
For example, when you select metric ‘Call Volume’ from the Key Drivers tab and select the dimension filters from the Add Filters. The Add Filters display only those dimensions filters that are associated with the metric Call Volume.
Also, the Add Factor list displays only those dimensions that are associated with the metric Call Volume.
If you select the factors Age, Area, Call Sequence, Call Status and City and change the metric from Call Volume to Call Goals.
You can see all the factors that are selected for metrics Call Volume, and when you hover the cursor over these factors a tooltip is displayed User Selection or Not Applicable and will not be used.
User Selection are those factors that are associated with the metrics Call Goals.
Not Applicable and will not be used are those factors that are not associated with the metrics Call Goals.
Insights - Ability to manage algorithms and hyperparameters for anomalies and key drivers
Now, WhizAI gives the ability to select algorithms and hyperparameters for anomalies and key drivers.
Selecting algorithm and hyperparameters for anomalies:
To select algorithms and hyperparameters for anomalies:
Go to Insight Workbench > Anomalies > Advanced.
Selecting algorithm and hyper parameters for Key Drivers:
To select algorithm and hyper parameters for key drivers:
Go to the Insight Workbench > Key Drivers > Advanced.
Insights - Enhancing Key Drivers analysis output
Now, when you perform key drivers analysis, the output is displayed separately for dimensions and metrics.
The result is displayed in two distinct tabs:
Contributors: It displays the result for dimensions.
Drivers: It displays the result for metrics.
To view the result of your metrics and dimensions separately from the Insight Workbench interface:
Go to the Profile icon > Insight Workbench.
By default, the Anomalies tab opens. Click Key Drivers and fill in the details in the columns of this tab.
Metric: From this column add Data Model and Metric.
Filters: In this column add filters to be applied on the scope.
Factors: In this column add factors to be used for potential key driver analysis.
After you click Analyze, by default the Contributors tab opens, where you can see the result of dimensions.
The left-side view of the Contributors displays the average contribution of the top contributor from each chosen dimension together with the total average.
To know more, click on the respective bar of a given dimension.
The right-side view of the Contributors displays the average, percentage contribution, and total volume for each of the dimension members.
You can hover the cursor over the bar to see the comparison.
Click the Drivers tab to see the result for metrics.
The left side view of the Drivers displays top driving factors for each metric.
The right-side view of the Drivers displays a scatter plot chart with the fitted line.
Using 'Day' expression in queries
We have enabled support for time granularity ‘Day’.
Similar to other time granularities such as year, quarter, month, week etc., you can now use the following time expressions in your queries.
Day over day (dod)
last <N> days
previous <N> days.
Refer to the following example queries supporting the above mentioned expressions.
Example NLQs with the responses:
Example 1: Show me sales for last 50 days
Example 2: Show me dod query count
Queries where the expression ‘Last/Previous N days’ is used with other time expressions are not supported. Examples: Last 10 days of 2020, Last 15 days of Q1 2021, Previous 25 days of H1 2021.
Co-referencing: Asking follow up queries with reference to the previous query
Now, you can ask follow-up queries by co-referencing the entities in the previous query. For example, you can use the following type of pronouns for co-referencing.
For example: If you have asked ‘Show me top 4 brands of last year’, you can ask a follow-up query as ‘How are they trending this year?’ WhizAI understands this follow up query and co-refers the word ‘they’ with ‘brands’ asked in the previous query.
First query: Show me top 4 brands of last year
Follow-up query: How are they trending this year?
For more information, refer to the following examples of follow up queries.
First query: Show me top 3 Regions in last quarter
Follow-up query: How were they trending last year?
First query: Show me top 10 accounts of last year
Follow-up query: Monthly sales for these accounts for current year
Co-referncing is not supported where the follow-up query includes a different metric than the previous query.
Co-referncing is not supported when the follow-up query includes a different entity from the dimension in the 1st (previous) query. For example: if first query has entity ‘Boston MA’ and follow-up query has entity ‘Chicago’,
Using ‘My’ expression in queries
We have enabled support for NLQs that contain the expression 'my'. You can use ‘my’ expression in your queries in following scenarios:
You can add and configure a data model, where you (user) are part of the source data. For more information, refer to the following example of source data.
Source data example:
Example NLQs: 'Show me invoices approved by me', 'Show me invoices assigned to me', 'Show me invoices where I am the approver' etc.
You can add and configure a data model, where entities (product, region etc) from source data are mapped to you (to user). For more information, refer to the following example of source data.
Source data example:
Example NLQs: 'Show me sales trend in my region', 'Show me TRx for my product', 'How is my product growth for last month?', 'Show my top performing product’ etc.
Example query 1: Show me TRx for my product
Example query 2: What is the sales trend in my region?
Supporting Word Sense Disambiguation
Now, WhizAI supports Word Sense Disambiguation. WhizAI's NLP engine understands the meaning of a word by identifying the use of that word in a particular context.
Consider following two queries for example:
Here in the 1st query the sense (meaning) of the word ‘Evolution’ is an entity name, whereas in the 2nd query word 'Evolution' is used as an intent (trend of TRx).
Introducing ‘Smart Search’: Search and add entities to queries
We have introduced Smart Search on the conversation box on the WhizAI Explorer. Smart Search allows you to search and add entities to your queries.
Configuring dimensions for Smart Search
From the admin console, you can choose the dimension to be enabled for Smart Search.
To configure the dimension for smart search:
Go to WhizAI Admin console > Content Manager > Configurations.
From the dropdown, select the data model for which you want to configure the dimensions.
Go to Dimension Selection For Smart Search and select the desired dimensions from the dropdown list. These selected dimensions appear in the Smart Search on the Explorer.
Click the Save option on the top right corner of the page.
To view these configured dimensions, go to Explorer > Conversation box > Smart Search, and click the select icon.
Configured dimensions are displayed as shown in the following figure.
You can search entities from these configured dimensions and add them to your query. For more information, refer to the ‘Searching and adding entities to your query’ section.
Selecting default dimension for Smart Search
You can select the default dimension to be available for Smart Search.
To set or change the default dimension:
From the WhizAI Admin console go to Content Manager > Configurations.
From the dropdown, select the data model for which you want to select the default dimension.
Go to the Smart Search Default Dimension, and select the desired dimension from the dropdown list.
Click the Save option on the top right corner of the page.
The selected dimension is set as the default search dimension for the Smart Search on the Explorer.
Searching and adding entities to your query:
Go to the Workspace > Conversation box > Search and enter the entity name.
As you start typing; the search result displays all the entities for the selected dimension.
Select the desired entity from the list; and the selected entity gets added to the query.
Click Enter; WhizAI displays the response to your query and the selected entity gets added to the context. You can continue to ask questions around this added entity.
Data Modeler: Viewing the data connection details of a pipeline
Now, on the Pipeline Manager user interface, you can view the data connection and the active data sources that are selected for a pipeline.
To view the data connection and the data source for a pipeline:
From the Admin console go to Data Modeler > Pipeline Manager.
Go to the Pipeline Name column and click the Connection Details icon as shown in the following figure.
Data connection and the data source(s) are displayed as shown in the following figure.
Data Modeler: Viewing the last 'Updated By' user, and the 'Last modified' date and time for the scripts
On the Script Editor user interface, we have added following columns:
In the 'Updated By' column, you can view the name of the user who updated/modified the script.
'Last Modified' column displays the date and time when the script was modified.
You can also sort the list of scripts by the last modified user or by the last modified date/time.
Data Modeler: Pipeline Manager user interface improvement
On the Pipeline Manager user interface, the Pipeline Name column is pinned to the left side. Even if you scroll to the right of this page, this column stays pinned to the left.
However, you can unpin this column, if required.
For more information, refer to the following figure.
Data Modeler: Viewing and editing the default configurations of the metrics
Introducing Configuration in the Data Modeler, where you can view/edit the default configurations of the metrics in a selected data model.
To view and edit the configurations of a metric:
Go to Admin console > Data Modeler > Configuration.
Select the data model from the Data Model drop-down list.
From the Value drop-down, select the metric(s) for which you want to edit the default configurations.
To edit the configuration, select the configuration as shown in the following figure.
Edit the selected configuration and click Save.
Following table explains different configurations that you can view and edit.
Data Modeler: New Setup (BETA)
Introducing New Setup in the Data Modeler. New Setup provides step by step guidance to bring the data model up in the system.
You can complete the data modeling workflow in following steps:
Adding a data model
Establishing the connection to the source data
Adding and configuring a pipeline
Running this configured pipeline for data ingestion.
For more information on data modeling workflow, refer to the WhizAI Admin manual.
In this version of the New Setup, multiple data connections cannot be established for a pipeline.
Data Modeler: Migrating pipeline configurations within environments
Now, you can migrate the pipelines from one WhizAI environment to another WhizAI environment. For example, you can migrate pipelines from the ‘Development’ environment to ‘QA’ environment, or from ‘QA’ environment to ‘Production’ environment etc.
To migrate the pipelines, you have to export the pipelines from the source environment and then import these exported pipelines to the target environment.
To migrate a pipeline:
Login to the source environment from where you want to migrate a pipeline.
From the Admin console go to the Data Modeler > Pipeline Manager.
Click the pipeline menu option at the right.
Click Export; the ‘Export Pipeline Configuration’ dialog opens.
From the drop down list, select the pipeline to be exported, and click Download to download a JSON file for the selected pipeline.
Login to the target environment where you want to import the exported pipeline.
Go to the Pipeline Manager user interface.
Click the pipeline menu option at the right.
Click Import; the ‘Import Pipeline Configuration’ dialog opens.
Drag and drop the configuration JSON file exported from the source environment and click Import.
Click Browse files to select the JSON file and then click Import.
Imported pipeline is added to the List of Pipelines page as shown in following figure.
Creating enhanced narratives using augmented functions in custom templates
Now, you can use the following augmented functions in your custom template to create enhanced narratives.
YTD: You can use the Year Till Date (YTD) function to get the value of the metric for the period starting from the start of the year till the current date.
Contribution: You can use the contribution function to get the percentage value of max. or min. contributing entities (e.g-product, region etc) from the provided response.
NxN: Use this function to achieve Period over Period (PoP) comparison for the given metric. For example, 4x4 expression compares the metric (or dimension) data for the current four weeks with the previous four weeks.
For more information, refer to the WhizAI NLP guide.
Customizing narratives displayed on a specific card on a pinboard
Now, as an Admin user, WhizAI allows you to customize the narratives shown on individual cards.
To customize the narrative from the card:
Go to the card on the pinboard and click the Narratives icon .
Narrative is displayed as shown in the following figure.
Hover the cursor on the narrative; Edit option displays.
Click the Edit icon.
The custom template that triggered the custom narrative opens as shown in the following figure.
If the card has Auto generated narrative, and you click the Edit icon; a blank narrative template opens and you can create a new card level custom narrative using this template.
Edit the narrative, as required, and click Save.
Go back to the card and open the narrative. The updated Narrative is displayed as shown in the following figure.
Click Save to save the card change.
Now, this updated narrative is attached to this card only.
If you save a card level narrative template in ‘Draft’ status, narrative will not be displayed on the card. It is recommended that after you edit/add narrative in the template and save it.
Was this article helpful?
Thank you for your feedback
Sorry! We couldn't be helpful
Thank you for your feedback
We appreciate your effort and will try to fix the article