Use Case:
We’re exploring the use of Pega Blueprint and Predictable AI to build a user-friendly process where a user can create an expense order by simply uploading an invoice in a conversational interface.
What We Want to Achieve:
We want the Conversational Agent to:
Accept an invoice document upload,
Understand the user’s intent (“Create Expense Order”),
Seamlessly trigger an Automation Agent to perform document extraction using Pega’s IDP or integrated OCR,
Auto-populate the extracted data into the ExpenseOrder case, and
Confirm the result back in the chat thread for user validation.
Current Gaps / Clarification Needed:
Is this end-to-end experience available out-of-the-box in Pega Blueprint (Conversational → Automation → Document Extraction → Case)?
If not fully available, what are the recommended components to be wired manually?
Can this integration be templated as a reusable pattern in future Blueprint releases (e.g., for expense processing, claims intake, onboarding, etc.)?
Enhancement Suggestion:
Introduce a Blueprint template or pattern where:
A Conversational Agent can directly invoke an Automation Agent, especially for document-based workflows.
Document upload and data extraction is a plug-and-play step.
Include default mapping for common document types (Invoices, Receipts, Purchase Orders) to speed up implementation.
Benefit:
This would significantly reduce time-to-value for common automation scenarios and enhance the composability and reusability of Predictable AI agents across business domains.
Looking forward to your guidance or roadmap input on this feature.
I’d like to answer your question as best I can - although I have made some assumptions about your use case in the process of answering. Your feedback is very helpful and gives me a lot to think about for the future of Blueprint!
Current Gaps / Clarification Needed:
Is this end-to-end experience available out-of-the-box in Pega Blueprint (Conversational → Automation → Document Extraction → Case)?
Blueprint can model the generative AI steps needed to structure the workflow you’ve described above. You would use a Generative AI step to represent steps 2 and 3. However, this is not something that would be configured OOTB upon importing your blueprint to Pega Platform. I believe those steps and the deployment of an agent to extract the data from the uploaded document would require some configuration in Dev Studio.
If not fully available, what are the recommended components to be wired manually?
Steps 3 and 4 would require configuration after importing the Blueprint to Pega Platform.
Can this integration be templated as a reusable pattern in future Blueprint releases (e.g., for expense processing, claims intake, onboarding, etc.)?
You can reuse previously created Blueprints to create new iterations of similar processes. I leverage the “Duplicate” feature in the Blueprint Dashboard to make modifications to a templated application.
I’d like to add a few points to what Casey mentioned.
We’re working on features that will enable you to have an Application-level Agent for your application when you import your Blueprint. This Agent would be smart enough to understand user intent (“Create Expense Order”) and trigger the corresponding Case.
As Casey mentioned, you can already configure a Generative AI step in your Case Lifecycle. However, this step needs to be configured further after importing the Blueprint in App Studio to process the Invoice document. This Doc AI feature will be available in Infinity 25.
@SumanKumar Thanks, Suman! Can we directly get functionality in Pega 25 without any further licensing rights or is it mandatory to get some license to use all these Agentic AI features?
Impressive integration approach automation flow feels much more refined this way.
Also exploring similar frameworks in public sector tools like BISP systems.
