Modern customer service—More social, more intelligent

RPA emerges for customer service

At the other end of the customer service spectrum is RPA. While social media responses are one-on-one, individualized interactions, RPA allows routine, predictable processes to be automated. This ensures that the interactions will be accurate and consistent. However, the boundaries of this technology are also changing, making an exact definition difficult. Increasingly it is encompassing more intelligence.

Although RPA has been around for decades, it has been getting more attention over the past few years. Originally, it was used mainly for back-office applications such as migrating data from one system to another. In many cases, integration of different systems was not feasible for technical reasons or because it would have been too expensive, so RPA was a good solution. More recently, customer-facing use cases have developed, both for external customers and for employees.

RPA is viewed in different ways, and beyond its basic use for simple repetitive tasks, there is some disagreement about the definition. “RPA is more of a continuum than a single solution,” commented Jim Sinur, a digital business consultant and CEO of Fleuresque. “At the low end, it carries out simple, point-to-point tasks and, at the high end, it becomes more of an intelligent bot that helps customer service agents or self-service customers.”

An example of a customer-facing RPA-powered bot is a stock trader’s (digital) workbench. “A trader handling a transaction manually needs to know what portfolio is being traded, rules for trading those items, and mathematical models,” said Sinur. “RPA can bring this information to one screen, avoiding ‘swivel chair integration’ in which the user must switch from one application to another to get access to all the required information.” RPA is often used in conjunction with chatbots to carry out processes after the chatbot determines the consumer’s intent.

RPA and the customer journey

Efficiency has been a big motivator for the use of RPA; in customer support applications, a goal has been to use the bot to help agents reduce handle time for customer calls. Now, the horizons are expanding. “The focus is on the end-to-end customer journey, during which we can automate the transitions from each channel to the next,” said Brad Beumer, customer experience and contact center lead at UiPath.

For example, the customer might start out on the website and find out more about the product through an automated chatbot session. At some point, their questions go outside the scope of the chatbot, and they get a live chat with a human, after which they might go to a store branch or place an order through the call center. “Each channel picks up where the last one left off so that the customer does not have to start over, allowing for a seamless experience for the customer,” said Beumer.

Chatbots are the user interface, and the robot works behind the scenes to deliver content or to complete a process. Some chatbots can ingest FAQs from a website, but if the answer is not there, a robot could call up content using an enterprise search engine. This pushes the initial boundaries of the definition of RPA. UiPath has sentiment analysis as part of its AI capabilities and can also integrate social media services. These functions can combine with the RPA to produce a wider range of responses.

UiPath works within numerous applications, including Amazon Web Services (AWS), Workday, and SAP. Its task-mining and process-mining features can identify opportunities for automation, and both human validation and built-in retraining can improve bot  performance over time. Its chatbots can discover customer intent, and then trigger RPA processes to resolve issues such as insurance claims by accessing back-office systems. “We provide excellent time to value so the benefits of automating the process can be realized as soon as possible,” noted Beumer.

Facilitating learning

Using RPA to speed up transactions is beneficial to customers and employees, and many companies want to launch such initiatives. Automation can take place more quickly if business analysts can implement it rather than requiring programming. Some RPA products offer a low-code or no-code capability, which allows non-programmers to develop RPA applications. The trade-off is that these options, particularly no-code approaches, provide less customization and scalability. However, they can get companies off to a quick start with automating processes and get the subject matter experts involved in the development phase. Even if no programming is required, IT should still be in the loop, since RPA generally requires touching enterprise systems.

The front end of an automated process is often a chatbot, with the RPA managing the process that the customer needs. “We automate both sides of the interaction,” said Kumaran Shanmuhan, chief growth officer at Jacada. “One side is the conversation that comes from the chatbot, and the other side is the process that is triggered by that conversation.” Jacada offers Agent Assist, which uses RPA to facilitate call center responses, and Customer Assist, which uses AI, natural language, and RPA to develop self-service applications. According to Jacada, Agent Assist reduces agent onboarding time by 75% and errors by 90%. Customer Assist has a library of prebuilt intents that is among the tools that allow business users to develop self-service applications.

“We put a lot of emphasis on how humans learn to do certain tasks and see how we can automate the consumption of knowledge,” continued Shanmuhan. “Knowledge is more than a series of keystrokes. It is therefore better to map the logic and thinking than to simply repeat the clickstream of an experienced agent,” he added. “For an agent just starting out, it helps when there is a system that understands natural language and presents relevant and actionable content. This improves customer service by providing agents with the right information when they need it.”