Process Intelligence at KofaxTransform

It’s after lunch on the second (last) day of Kofax Transform, and the bar for keeping my attention in a session has gone up somewhat. To that end, I’m in a session with Scott Opitz and Rich Rabin from the Kofax Altosoft division, but not sure it’s going to meet that bar since Opitz started out by stating that what the TotalAgility (KTA) sessions call process is a much more complex than what they call process, and I’m a bit more on KTA’s side of this definition.

Altosoft process intelligence is really about the simple milestone-based monitoring processes of operational intelligence, with the processes being executed on multiple systems, more like SAP’s SAP Operational Process Intelligence based on HANA or IBM Business Monitor; you rarely have all of your process milestones in a single system, and even if you do, that system may not have adequate operational intelligence capabilities. Instead, operational intelligence systems pick up the breadcrumbs left by the processes — such as events, database records or log files — and provide an analytics layer, usually after importing that data into a dedicated analytics datamart.

There are really two main things to measure with process intelligence: performance and quality/compliance. To get there, however, you need to know what the process is supposed to look like in order to measure patterns of behavior. Altosoft’s process intelligence does what they call “swimlane analysis” — looking at which tasks are done in which order, a form of process mining discovery algorithm since there is no a priori process model — to identify operational patterns and derive a process model from runtime data, showing the most common/expected paths as well as the outliers. Not just process mining as an analysis tool, it then shows the live process monitoring data points against those models, and provides some good interactive filtering capabilities, allowing you to find missing steps that may indicate that the task wasn’t performed or (more likely for steps with manual logging) that the task was not documented.

Since the Insight platform is a complete BI environment, this information can also be combined with more traditional BI analytics and dashboards, providing real-time alerts as well as historical analysis. They also have ways to use a predefined process model and measure against that; this then becomes more of a conformance analysis to see how closely the actual runtime data matches the a prioiri model.

Leave a Reply

Your email address will not be published. Required fields are marked *