Decision Management

What's new in Automation Decision Services - Cloud Pak for Automation v20.0.3

By Antony Viaud posted Fri December 18, 2020 04:01 AM


IBM Cloud Pak for Automation 20.0.3 has been released on December 18, 2020. For its Automation Decision Services component (ADS), version 20.0.3 is the second release, and it contains many exciting new features and enhancements to facilitate the creation and integration of decision automations. Let's review some of its new features:

Enhanced management of decision services, still with an intuitive touch
Collaborators with user roles and permissions can now be defined for decision automation projects, to control who can access and modify decision services.
In the updated terminology for Business Automation Studio and ADS, a decision project regroups one or many decision services and is attached to a remote Git repository.

Define collaborators and their permissions for a decision project

Decision Designer users can now view the history of a decision project and manage its versions (as tags in Git).

View history and manage versions of a decision project 

The out-of-the-box user experience is improved, with a new feature for users to build and deploy decision services from Decision Designer, using an embedded build service and runtime. This comes in addition to the ability for clients to use their own continuous integration and delivery tools.

Deploy any version of a decision service

The two ADS releases have already introduced many improvements for decision modeling and ADS can now import decision models created with the previous generation of decision modeling in IBM Operational Decision Manager or IBM Decision Composer.   
Easy creation and updates of intelligent decisions using ML predictions
Users can now easily update a predictive model in ADS when the associated ML model has been updated, for instance after the ML model has been retrained.

Update a predictive model based on new input parameters of the ML model

With the new creation process for predictive models, users can also start modeling a decision that includes predictive models, before ML models are available, allowing a variant of "top-down" modeling.
Note: for a more detailed description of how ADS 20.0.3 is facilitating updates to machine learning models, see this blog post.
ADS 20.0.3 also adds connectivity with AWS SageMaker through a new proxy provided as open source. Other new open source assets include notebooks for training and deploying ML models.
New capabilities to integrate and monitor runtime decision services
Cloud Pak for Automation 20.0.3 features a new automation service catalog that accelerates the creation of a business solution.
Decision services created with ADS can be published as automation services. Such services can then be easily discovered and consumed by workflow processes or applications built with Cloud Pak for Automation.
This new version of Cloud Pak for Automation also enables IT users to monitor runtime decisions thanks to Business Automation Insights (BAI) and its new Business Performance Center dashboard. When activated, ADS sends execution details for each decision, which enables BAI to process all that data and enable users to create meaningful dashboards.

Runtime decision data in Business Performance Center