The journey to intelligent automation starts with RPA
Automating higher-order tasks
Robotic process automation (RPA) represents a strategic investment in intelligent automation. It combines RPA with machine learning and artificial intelligence to automate higher-order tasks that otherwise require human perception and judgment. The intelligent automation journey improves automation from a rules-based approach with structured data, to a data-driven approach with unstructured data. As a result, automation evolves from "doing" (RPA) to "learning" (ML) and eventually to "thinking" (AI).
From rules-based to learning-based automation
Initially, intelligent automation will combine AI with cognitive capabilities to improve and extend RPA automation. In 2020, UiPath and other RPA vendors released exciting cognitive capabilities, including computer vision and natural language processing. Incorporating algorithms and various cognitive technologies will improve accuracy of complex business processes that involve unstructured information.
The impact of intelligent automation will be three-fold. First, automation will extend to unstructured data in text as well as speech. Examples include financial documents, emails, and audio conversations. Next, automation will enable longer, more complex workflows that incorporate business decisions and exceptions not based on rules or structured data. Examples include incorporating confidence ratings to decide whether a decision is automated or escalated to a human. Finally, flexible and robust automation will be implemented that are capable of "reading" and "recognizing" user interfaces without complex programming.
To illustrate the power of cognitive capabilities, we'll examine how a financial document is automated with RPA and IA, respectively. While the US individual tax return Form 1040 is a structured document, supplier invoices may be unstructured documents.
For example, invoices from different suppliers contain different data, use different formats, and may be different file types. With RPA, each supplier invoice type requires a separate workflow, activities and technologies to read, extract and manage invoice data. In addition, both unattended and attended automation may be necessary to enable a human to review and decide certain tasks. On the other hand, IA would leverage the same, standardized workflow, activities and technologies across all supplier invoices. While it may still be necessary to have a human in the loop, the attended tasks would be minimized and based on confidence ratings.
Customers would be unable to efficiently and effectively automate invoices from hundreds of suppliers using only RPA without cognitive capabilities. However, IA scales and extends invoice automation. There is no need to build and maintain hundreds of different RPA workflows. Organizations will be able to build standardized workflows that should enable all invoice types. As this example illustrates, the benefits of IA are greater than scalability and extensibility. Additional benefits include customer experience, accuracy and compliance, cost, and ultimately, time to value.
Remember, the journey to intelligent automation starts with RPA!