AI-based Business Process Automation in the Enterprise

by Erhan Cakmak

Summary

The fourth industrial evolution—the melding of the cyber with the physical—is well underway. Industry 4.0 is impacting not only Operational Technology, but Information Technology as well. This can most readily be seen, perhaps, when one considers how machine learning and artificial intelligence is driving efficiencies in business processes that begin with physical documents, digitize them, and then classify, enrich and dispatch them to workflows before they are, finally, archived in document management systems.

Embedded “Digital”

“Digital” is now firmly embedded in every business. But even with technology as an integral part of the organization and its strategy, it is people who will ensure success in a world that continues to reinvent itself at an unprecedented rate. Simply adding more technology to the enterprise is insufficient; we must focus instead on enabling people to do more with that technology.

Intelligent Business Process Automation is the most promising of all symbiotic progressions. Intelligent automation is the launching pad for new growth and innovation, premised on the harnessing, by people, of the modern machine. Powered by artificial intelligence (AI), the next wave of business automation solutions will gather unprecedented amounts of data from disparate systems and—by weaving systems, data, and people together—create solutions that fundamentally change the organization, as well as what it does and how it does it.

AI will have a major impact on the world of work. It can complement human input in complex work requiring creativity and judgement, and will increasingly act as a substitute for routine labor. Task assistants and associate systems will allow humans to delegate work to a computer with the appropriate interfaces.

Automation improves process efficiency—business process management and rule engine software are the basis of most current process management systems in enterprises, such as Pegasystems and SAP Ariba.

Systems such as these enable enterprises to model their processes, with computerized orchestration of the decisions on where to direct flows downstream. These systems are already powerful in the industrialization of processes and in managing business rules such as policies.

However, they are configured systems and extremely systematic and therefore unable to adapt to any change. They work for what they are designed for. Any situation not accounted for during the configuration phase is an exception requiring human intervention.

In essence, these business process management systems deliver process efficiency against the designed and established process. Cost savings are achieved by accelerating the process through computerized execution of the associated tasks.

Further, existing systems tend to have one dimensional, simplistic classification abilities upon document ingestion. With classification so poor, documents tend to go through the most complex workflow, with the most touch points, whether or not it is needed, wasting resources.

Extract, Classify, Process and Visualize

In contrast, AI-based business process automation helps processes become increasingly intelligent. AI-enabled processes adapt to change and, over time, become more precise. As the quantity of data processed reaches critical mass, human intervention is only in very rare and exceptional situations.

AI-based automation independently maximizes its efficiency simply by the way it operates—learning by “doing.” It automatically learns from changes and promptly reacts to change without any rule set, minimizing operational efforts. Rule-base systems quickly reach their technical limits; they cannot reach the full potential of automation. The only way to process unstructured data efficiently is by using computer linguistics and artificial intelligence to analyze and classify raw documents at the source.

Document classification can now be more sophisticated, and adaptive. More granular, automated, classification—with better enrichment too—at ingestion directs documents to the most appropriate workflow, rather than the “catch all” workflow. Processing times are shortened not just through computer execution but also by mapping the documents to the correct process. And machine learning can automatically identify new classifications of documents worthy of their own, new, workflows. And it’s not just for “new” documents—databases of existing electronic documents can be processed as well for classification, enrichment and training the system.

At the back end, large organizations rely on small army of analysts to execute reports on an ad hoc basis, sifting through the output of the workflows. Even with modern visualization tools, many “sources of truth” exist throughout the organization as data is spun out from central repositories to individual work groups, the members of which, knowingly or not, apply their own biases to their analyses, creating their own “truth.” Business leaders are increasingly realizing the value of visualization platforms that analyze data in situ, from the central repository, in real time.

This new approach is already finding application in large organizations at the intersection of electronic content management and customer experience management. Technology solutions now encompass all customer facing interactions and connect those with the relevant business content. Inbound customer correspondence is classified by content and prioritized by urgency, then handed over to back office systems without human intervention. Semantic algorithms automatically analyze documents to extract case data and create value-added datasets.

In such organizations, a culture shift is underway as business leaders place increased emphasis on the value of customer experience management, while at the same time realizing that the multidimensional, often messy, business of customer relations can be tamed with the assistance of technology. Processing times are shorter, the quality of the customer experience is higher and labor costs are lower.

Fit and Needs Assessment for AI-based BPA

The frequency with which organizations must reinvent themselves is increasing. AI-based business process automation is a key enabler of dynamic responses to market disruptions and competitive pressure. Deployed adroitly, it complements the workforce and works hand-in-glove with it. Business leaders in search of an application strategy can start with four simple steps:

• • Identify discrete business processes or workflows that require frequent and manual updates, rapid scaling, data extracts, and/or a high degree of personalization. Any application that is heavily reliant on data is a top candidate for self-evolution through machine learning.
• • Take an inventory of labor-intensive business processes and identify appropriate opportunities to invest in automation and machine-learning capabilities. See where these intersect with data-reliant processes and further prioritize accordingly.
• • Map documented AI-based BPA success stories against current business processes to prioritize specific opportunities—align proven use cases to your own.
• • Map the implications of tasks being automated—the changes to roles, organization needs, processes and skills.

Benefits

• • Lower labor costs
• • Faster processing times
• • More efficient processes
• • Higher quality customer experiences

Conclusion

Adaptive business processes are mandatory for the digital age enterprise. Such processes are enabled by intelligent business automation technologies, which, complementing the human workforce, drive labor and other efficiencies while encoding organizational DNA that self-learns and evolves. Even existing processes can benefit from “bookends” of better, more granular, front-end data classification and single-source-of-truth visualization of the outputs. Proactive enterprises are currently plotting their strategies to harness the symbiosis of intelligent automation and human agility.