Digital Duo Could Take Process Automation to the Next Level

Robotic process automation (RPA) and machine learning (ML) are two technologies that figure largely in the digital transformation of supply chains. In combination, they are creating new possibilities for automating processes.

Co-authored with Ken Cottrill

RPA, probably the lesser-known of the two, involves software rather than physical robots and incorporates applications for processing transactions. A simple example is automating a response to an email. More complex operations could be customer relationship management responses, orchestrating trade finance documentation, or updating procurement records.

When combined with ML — systems that automatically learn and improve without being programmed — RPA can automate more complex transactions.

Features of robotic process automation

A group of MIT CTL Exchange partner companies explored applications of RPA in the context of supply chain digital transformation at the Digital Transformation in the Supply Chain Roundtable, February 26–27, last year[i]. An MIT CTL roundtable/symposium on data analytics scheduled to take place this March will explore the topic further (see below).

One of the virtues of RPA is that its implementation is relatively straightforward.

Participants at last year’s roundtable agreed that it’s not necessary to hire experts to develop RPA solutions since companies can teach employees to build the requisite software capabilities. At one company, a two-week training period was sufficient to create an RPA application. Moreover, employees are much more likely to know whether a problem is worth solving using RPA than an external consultant. Companies can train “superusers” who bring the technology to their departments and the broader organization. One roundtable participant has trained 200 people in this way.

A surprising feature of RPAs is the need to involve human resources in their deployment. A leading user of the technology at the 2019 roundtable explained that an RPA bot is a virtual employee in at least three ways. First, the bot needs a computer account, password, and appropriate access privileges. Second, it reports to someone responsible for monitoring and managing the application. If that manager leaves or moves to another position, the oversight responsibility needs to be transferred to another person. Thirdly, if the bot takes over a substantive part of another employee’s duties, HR may need to reassign the employee or retrain him/her to do other tasks.

Powerful partnership for enterprise

It is relatively easy to bolt RPA on to enterprise applications. For example, a leading electronics company at the digital transformation roundtable connected an RPA solution to an optical character recognition function. Consequently, marrying RPA with ML is fairly straightforward technically, yet the combination can be extremely powerful.

As the Wall Street Journal reported recently, current examples of these combinations include projects to verify signatures on checks, assess insurance claims, and detect fraud in paperwork. Automating these processes is a complicated challenge that requires the learning capabilities of ML. For example, signatures on checks can vary in ways that are difficult for automated systems to recognize and validate, reports the Journal. In early testing of the ROI/ML signature verification application, the bots achieved a success rate of 80%. The 20% of checks that were not cleared with absolute certainty by the solution were returned to human tellers for final verification.

Fraud detection is a promising supply chain application of RPA/ML solutions, especially where complicated documentation is involved. Supply chain planning is a natural application area for these solutions. In a process characterized by manually intensive and repetitive tasks, bots can be used to calculate and update well-structured forecast models covering hundreds of thousands of items. Moreover, companies can automatically purchase the items regularly with no human intervention using RPA technologies.

Future growth of robotic process automation

RPA is the fastest-growing segment of the global enterprise software market, according to research firm Gartner Inc., and spending on the software is expected to reach $2.4 billion in 2022.

Solutions that marry RPA with ML will play an increasingly important role in many industries, including in supply chain management.

[i] Digital Transformation in the Supply Chain roundtable, February 26–27, 2019, hosted by Dell, Austin, TX.

The Analytics & Automation of the Future — North America symposium/roundtable will take place from March 10 to March 11, 2020, at Dell Technologies, Franklin, MA. More information is available here.

The event will also take place in Singapore from March 24 to March 25, 2020. It is the latest in a series of events convened by MIT CTL to explore different aspects of machine learning. More information is available here.