robot typing on laptop (Zapp2Photo/

Building a better chatbot

Chatbots are showing up on websites of city halls and state legislatures to answer common questions, often serving citizens faster and offloading some customer service tasks from public sector employees. But chatbots might actually prove more useful for agencies' internal applications, especially when combined with robotic process automation, according to a new report from Forrester.

Releasing a chatbot into the wild -- or putting it on a citizen-facing website -- means it will encounter a broad array of questions from a variety of people who all communicate differently. The knowledge engines that drive today's chatbots just aren’t advanced enough to handle this diversity very well, according to Craig Le Clair, an analyst at Forrester and the author of the report.

With a chatbot designed for a specific agency use, “the context of the conversation is much more succinct -- it's much more well defined,” Le Clair said.

Additionally, employees can be trained how to best use an agency chatbot, said Accenture Federal Services CTO Dominic Delmolino, whereas citizens likely have little idea what the bot can and can't deliver.

Chatbots can be even more useful when they're connected to robotic process automation (RPA), which automates repetitive tasks, like moving files or filling out forms. It is informed by a process programmed by humans.

RPA-powered chatbots can be used for IT support, so an employee who needs a password reset can access potential solutions without taking time from a network admin. Similarly, a human resources application could help staff members update their personal information on their own, Le Clair said.

RPA is most useful for automating highly repetitive tasks that occur at a high enough volume within an organization to justify the implementation cost of automation that can be in the six figures, Delmolino said.

Government has already showed interest in RPA. NASA has its own digital employee, Washington bot, which was built to copy information from emails that are sent to its inbox and populate files in an HR case management application. The Treasury Department’s Bureau of the Fiscal Service has launched an RPA pilot to streamline common financial management processes. Already, the bureau learned that up to 80 percent of the process steps that are part of the pilot can be automated.

Another potential use for RPA will be as a way of web-enabling a legacy system. The legacy system would be wrapped in an RPA that acts as an application programming interface for the legacy system, Delmolino said.

“It might be possible to … front the legacy system with an RPA bot, and that provides the interface and it does the keystroking and what’s necessary to interact with the legacy system,” he explained.

One of the advances coming down the line for RPA will be the integration of more intelligence, both Delmolino and Le Clair predicted.

“It’s pretty dumb, really, today,” Le Clair said of RPA. But as machine learning becomes integrated with the technology, RPA will begin making more decisions independent of human involvement, he predicted.

Some of this integration of AI into RPA is already happening, according to Delmolino. An RPA for the insurance industry uses image recognition technology on photos submitted by customers to make decisions about claims, he said.

Going forward, RPA systems must also be more easily auditable, Delmolino said. Right now, while the decisions are mostly programmed by humans, it’s all in computer code. That means if the legal department needs to understand why an RPA made a particular decisions, the code would need to be translated by an engineer. This translation and justification for the decisions being made should be more clearly defined within RPAs, he said.

For all their benefits, RPA represent a nascent market. ”It’s pretty low level of maturity right now,” Le Clair said about the state of RPA saturation. “You have early-stage deployments at larger companies.”

Although private- and public-sector enterprises are concerned that the efficiency-boosting technology could eliminate jobs, Delmolino said doesn’t see that happening. Nevertheless, the government wants to look at it closely before heavy implementation, he said.

The General Services Administration's Emerging Technology group announced it plans to hold an RPA event later this year to gauge agency interest on the topic.

About the Author

Matt Leonard is a reporter/producer at GCN.

Before joining GCN, Leonard worked as a local reporter for The Smithfield Times in southeastern Virginia. In his time there he wrote about town council meetings, local crime and what to do if a beaver dam floods your back yard. Over the last few years, he has spent time at The Commonwealth Times, The Denver Post and WTVR-CBS 6. He is a graduate of Virginia Commonwealth University, where he received the faculty award for print and online journalism.

Leonard can be contacted at or follow him on Twitter @Matt_Lnrd.

Click here for previous articles by Leonard.

inside gcn

  • health data

    Improving the VA patient journey with data transparency

Reader Comments

Thu, Apr 19, 2018 Jim Bard

Have you seen any good examples of governments using chatbots and RPA to streamline procurement process and help staff find goods / services through existing contracts?

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group