This first of two sessions highlights improvements to the Low Code and AI capabilities of Pega Platform.
Sam Tremlett (@trems1 - Senior Product Manager, Low Code, Pegasystems) shared enhancements to the low code authoring experience, focusing on App Studio and highlighted new features such as
Many to many relationships,
Native support for ODATA integration,
Contextual knowledge articles, and
Out of the box CRUD actions for data objects.
Andy Bober (@bobea - Senior Director, Product Management, Applied AI in Intelligent Automation, Pegasystems) told us about "AI for every user", showcasing how Pega has thoughtfully implemented AI throughout the Platform to make every user more productive, including:
Prediction widget & overlay
Prediction output label
Monitoring & Reporting improvements
In this post you will find:
The recording of the webinar (including the Q&A answered live during the Americas, EMEA & India) event;
Key Questions & Answers from both sessions in the event
CLSA Community - AI for Everyone & Low Code
Questions and Answers
Low Code and "Choose Best Ruleset"
Do we still have UI Pages in 8.8?
The Portal authoring experineces in Constellation and Theme Cosmos offer the ability to create landing pages for the application.
How can App Studio developers make sure that app studio development generated rules are being saved into right ruleset?
This is defined by the Choose-Best-Ruleset algorithim which is used by both App Studio and Dev Studio.
This algorithm uses the application stack, open rule sets, rule set of the classes of the rules and branch preferences. Using Branch Development preferences, the App Studio developer can control the application layer a rule will be placed in.
If rule exists it will attempt to update in place
A new rule will be created in the ruleset where the Class containing the rule is defined
If previous rulesets are not valid, will attempt the highest ruleset in a layer.
This allows a user working in implementation App to make changes to a build on framework app.
We have technical materials that outline the process if details on Choose-Best-Ruleset are needed.
Can you tear-off the Assistant panel?
The panel follows the right hand slide out UX pattern and cannot be "torn-off" into a separate window.
Is the Developer Assistant available in Dev Studio as well?
Developer Assistant is currently only available in App Studio, but I would expect us to extend this to Dev Studio in future releases.
Can we extend the Developer Assistant tool to use in Portals, like a Customer Assistant? For example: to give users contextual knowledge articles while working on a case.
The extension guide for Developer Assistant panel is included in the 8.8 release.
Is the application security checklist available to extend as all might not be relevant to our applications?
The Application Security Checklist is a final rule, as they represent the minimum security checks that should be checked for every app.
You can author your own application guides.
Many to many relationships
Is the "index class" created for the Many-to-Many relationship for better reporting?
It does allow for efficient reporting, however the main objective is to avoid locking and the need to keep data synchronised across multiple object instances.
The index table Key is a GUID, but
The To ID and From ID are optimized for reporting.
Does the Many-to-Many data relationship create any Data Pages in backend to fetch data?
Yes, several data pages are created depending on the configuration. Data pages are created to get an instance and to get a list of instances given a To ID or a From ID.
Use cases that data pages address:
Get list of items related to me
Open a specific relationship
Is there be any difference between REST and ODATA configurations?
If I already have some service configured with REST can I change to ODATA?
OData is a set of standards in implementing REST protocols so the changes required will depend on the vendor who provides the endpoints.
It is possible that you would not need any changes to an existing implementation but if you do, the wizard will cut the time required to do so from hours to minutes.
Is ODATA integration supported from Dev Studio as well as App Studio?
Supported in both, but the integration wizard is only available in App Studio. This means if building an ODATA integration in Dev Studio, you will need to author on a rule by rule basis.
Does ODATA configuration also support Application settings?
Yes, application settings are supported in ODATA as well.
CRUD data object management
Are the CRUD operations using REST in the background?
CRUD operations in the Constellation applications are using the Constellation API (DX API).
Classes inheriting from Work- will use the /cases endpoints;
Classes inheriting from Data- will use the /data_views endpoints.
Is RBAC/ABAC used to provide access control on data objects accessible via the CRUD capability?
We are generating a RBAC rule on the backend. You can open this rule in Dev Studio if you need to add more granular access controls like conditional class access. The experience in App Studio allows you to control the "large bucket" access like read/write/delete.
This is a great example of how both App Studio and Dev Studio can be used in tandem when the complexity of your use case requires.
AI for Everyone
Is Process AI already integrated with Pega Platform v8.8 or is it still a separate license?
While Process API is fully integrated in the platform, it does require a special license.
What ground work is required to be done by an LSA/SSA before handing over to citizen developers/BAs to design forms, process etc in App Studio?
Most of the "setup" for Process AI in a case can be performed directly in App Studio and Prediction Studio. Of course there will always be tasks that require LSA/SSA guidance, such as configuring integrations with data sources.
When the same case is dealing with multiple predictions, how does Process AI manage the conflicting predictions for the case?
There can be multiple predictions in the same case, each predicting different outcomes. These are not conflicts.
Depending on the use case, one can also call on a single prediction several times during the lifetime of a case, the likelihood of an outcome can change over the lifetime of the case as more case data is collected and as the prediction model learns more over time.
What kind of criteria is needed to enable the "Notification" option within the Model view?
Is Process AI different from Process Mining in Pega?
They are different. Both these solutions work towards optimizing processes and reinforce each other. Pega’s process mining capabilities come from the acquisition earlier this year of Everflow, which is a powerful and broad solution for process mining. We have an early adopter program for Pega clients who run a Center of Excellence. We are planning to provide process mining in an upcoming release.
Adaptive models used to learn in real time, updating when input data updated. Why are we adding a pooling interval for adaptive models?
Adaptive models are still updated continuously. The trigger for an update will now be time-based, which is a more safe and robust mechanism and reduces the need for each application to define the update threshold.
From QA standpoint, how QA verifies the predictive models as they are adaptive. How system derives the QA expected outcomes?
In a production environment, the adaptive models will adapt their behavior as new evidence is being recorded.
In other environments, QA can be set up with expected results.