0 votes
20 views
ago by (30.6k points)
IBM markets CP4BA as "AI-powered automation". For a BAW team that is a vague statement. Which concrete AI features exist in the Cloud Pak, which ones are usable from a workflow today, and which require additional products or licences?

1 Answer

0 votes
ago by (30.6k points)

The AI in CP4BA is a set of separately deployable capabilities plus generative AI features inside Workflow itself. What each one does and how a BAW process uses it:

CapabilityWhat it doesFrom a BAW processNeeds
Generative AI activities in Workflow (CP4BA 24.0.1+, BAW 25.x)text generation, summarisation, extraction and classification steps in service flows, backed by watsonx.ai foundation modelsa generative AI task in a service flow, prompt with process data, output mapped to variableswatsonx.ai (IBM Cloud, or on Cloud Pak for Data from 25.0), API key, egress policy
Workflow Assistant in Workplace (25.0.x)conversational help for end users: find and filter tasks, summarise a task, generate e-mails / commentsno change to the process; benefits from good task subjects and business datawatsonx.ai provider configured
Automation Document Processing (ADP)classification and data extraction from documents with deep learning models trained in Document Processing Designer, with verification UIsdocument processing activities in a case / workflow solution; extracted fields become process datathe document_processing pattern (capacity: GPU optional)
Automation Decision Services (ADS) with predictive modelsrules combined with machine learning predictions (models from Watson Machine Learning or an Open Prediction Service)call the decision service from a service flow (discovered in Studio)decisions pattern, a ML provider
Business Automation Insights (BAI)event streaming and dashboards over process data; the data source for ML on process behaviourautomatic once enabled; feeds Process Mining and custom modelsbai component, Kafka / OpenSearch
IBM Process Miningdiscovers the real process from event logs, finds bottlenecks and deviations, simulates changes, exports BPMNBAW instance / task data exported to Process Mining; BPMN skeletons imported into Workflow Centerseparate deployment (included in the CP4BA entitlement)
IBM RPAbots with OCR / document and text AI for legacy UIsbots invoked from service flowsRPA server
watsonx Orchestratean AI assistant / agent platform that discovers BAW processes and services as skillsexposed processes become skills; newer releases add Model Context Protocol tooling for agentswatsonx Orchestrate (separate product)

Where to start as a workflow team: the generative AI activity (summarise a claim, draft a response, classify a request) gives visible results in a week and needs only a watsonx.ai project; document extraction with ADP is the next biggest lever for document-heavy processes; predictive decisions in ADS and Process Mining pay off once BAI data has accumulated. Everything AI-generated should pass through a human task when it affects a customer (question on human-in-the-loop design).

References

Related questions

723 questions

807 answers

98 comments

4.8k users

Join BPM Community Discord Channel

Welcome to BPM Tips Q&A, Community wiki/forum where you can ask questions and receive answers from other IBM BPM experts and members of the community. Users with 2000 points will automatically be promoted to expert level.
Created by Dosvak LLC
Our Youtube Channel
...