Architecture and workflow
Event flow and data processing
Imagine a company using IBM Business Automation Workflow to manage customer onboarding. As customers submit forms, tasks are completed, and decisions are made, each action generates an event. Here's how Business Automation Insights processes those events:
- Event Emission The automation applications—such as Workflow, Case Manager, ODM, or RPA—are configured to emit events. These events describe what happened (for example., “Task Completed,” “Decision Made”) and include metadata like timestamps, user IDs, and business context.
- Streaming via Kafka The emitted events are sent to Kafka, a distributed event streaming platform. Kafka acts as a high-throughput channel that ensures events are reliably delivered to BAI.
- Data Processing BAI ingests the events from Kafka and processes them using its internal
pipeline. This includes:
- Filtering and transformation to standardize event formats
- Aggregation to compute metrics like average task duration or decision frequency
- Enrichment to add business context or correlate related events
- Storage in OpenSearch The processed data is stored in OpenSearch, a fast and scalable search engine. This allows users to query and retrieve metrics efficiently.
- Visualization in dashboards Business users access the Business Performance Center,
a no-code dashboard editor. Here, they can:
- Build visualizations like bar charts, line graphs, and KPIs
- Monitor real-time performance
- Share dashboards with stakeholders
To help illustrate this flow, the following figure shows how events move through the Business Automation Insights architecture—from emission to visualization.

Deployment and configuration
- Deployment
-
Business Automation Insights is typically deployed on Kubernetes. Installation involves configuring event sources, setting up processing pipelines, and connecting to OpenSearch for dashboard visualization. See Installing a Business Automation Insights production deployment, Configuring Business Automation Insights for event processing, and Architecture and workflow. You can customize your deployment as necessary.
- Event emission
- Events can be emitted from IBM automation products or custom applications. You can secure and configure this flow using the processing application. See Events from custom sources and the appropriate pages in Administering.
- Troubleshooting
-
If results are not as expected, use logging and tracing tools, check your configuration files, and validate component readiness. See Troubleshooting.
Samples and resources
For material to handle custom events, see the BAI emitter
samples
GitHub page.