Top considerations for ECM and content services
♦ Metadata: Metadata management is utilitarian for content intelligence and records management. It’s the basis for filing or storing content based on objective facts such as an NDA (non-disclosure agreement) or specific customer. This means that “the legal team can find it because it’s an agreement, the project team can find it because it’s associated with a project, and an account manager can find it because it’s associated with a customer,” Robertson explained. Metadata is also helpful for cataloging content. Analyzing and extracting relevant information from content “gives you that catalog which then allows you to build up that dictionary of metadata terms, which then become part of how your systems talk to each other,” said Chris Wynder, director of product marketing, Developer Cloud and OEM, at OpenText.
♦ Taxonomies: The value of taxonomies to content services is evolving. Certain AI approaches obviate taxonomies by simply learning the meaning of terms in a corpus, for example. However, “for existing users, their initial ability to get that automation is going to be low because it doesn’t know anything yet,” Wynder warned. There are also approaches in which users formally define a centralized taxonomy prior to implementing content intelligence—which can be time-consuming. The best option is likely a synthesis in which taxonomies aid machine learning so “if you want users to kind of reinforce that model, it has to be their taxonomy to use,” Robertson said.
Process automation provides the necessary action to capitalize on the aforementioned aspects of content intelligence within the broader context of business functions or workflows. There are several characteristics of the deterministic action required of process automation.
These include the following:
♦ Low-code/no-code design: No-code techniques permit citizen developers or end users to design applications and processes for workflows. Examples include demonstrating the actions that bots will duplicate to file a document, or using intuitive, visual methods to build applications for onboarding new employees. “No-code capabilities allow for approval processes and very easy-to-use drag-and-drop form building, and a canvas to lay out what the flow should be from person-to-person to plan when back-end system processes are needed,” Rapelje said.
♦ Orchestration: Runtime orchestration to coordinate efforts between systems or applications is an integral element of implementing processes across them while broadening the scope of content services deployments. “Robotic process automation is really about orchestration,” Kohli said. “It’s about initiating, monitoring, and running bots and bots made up of other bots.” Bots can orchestrate the discrete steps to reach adjudication in an auto insurance claim, for example.
♦ AI: There are few dimensions of content services not impacted by AI; process automation is one of the more conspicuous. Kohli described a healthcare use case in which bots equipped with deep neural networks for image recognition, as well as with natural language processing capabilities, are able to diagnose patients in various stages of terminal illnesses in accordance with information in their medical charts. “For the hospital to act on something like this, they need to be able to go into the hospital’s ERP [enterprise resource planning] systems and patient record systems,” Kohli specified. “RPA will run that entire end-to-end process, and, when needed, it will call the right image recognition AI and integrate that into the RPA process.”
Gaining productivity insight for an individual employee, business unit, or workflow is a crucial byproduct of process automation that is aligned with the analytics Reynolds referenced. Productivity intelligence into specific workflows such as invoice processing, for example, can be based on analysis of how often invoices are paid to whom, “and then you can create whatever charts and slides of the data that you want,” Kohli explained.
In fact, one of the chief advantages of process automation is deriving insight into how productive business units and processes are, along with “scale, reduction of errors, and the ability to monitor,” said Kohli. One concept that people forget is that when software is doing these things, it is possible to see what is working well, and what the status is—and do analytics on that also, Kohli added.
Companies and Suppliers Mentioned