At One Of The Biggest Health Insurers, A Digital Employee Winning Over Customers’ Minds


Is the “digital workforce” now a thing? One of the biggest health insurers and providers in North America, Humana, has accepted an intelligent virtual assistant as a full-fledged employee. This assistance is not viewed as a tool or a machine, and it is not replacing somebody. Instead, it’s a super-helpful “coworker” that develops and learns alongside the medical and insurance experts with whom it collaborates.

According to Joe Bechtel, principal of business automation strategy development for Humana, the “Allie” intelligent assistant was created to lessen the documentation burden experienced by the company’s clinicians. At the most recent Automation Anywhere conference in New York, I got the chance to meet with Bechtel. where he talked about creating Allie and introduced her to the staff of his business.


Clinicians’ administrative workloads have been cut by up to 15% as a result of Allie, allowing them to concentrate more intently on patient care, according to Bechtel. “We initially established a task force to aid in the enterprise-wide scaling of our machine learning and desktop office automation. We saw automation as a tool that would help us communicate with our employees more personally.

80% of clinician time was spent “documenting their encounters, through treatment planning and medical charting, rather than connecting with their patients,” according to Bechtel’s team as they dissected and examined their organization’s quality processes.

Allie’s success was largely due to the close involvement of clinicians in the procedure from the beginning. “Clinicians would always tell me, “You don’t understand my job,” whenever I would visit a business unit within the organization. You just don’t understand it,’ he says. “We address that by integrating people into the process, allowing them to be a part of that journey from ideation to design and implementation,” the author said.

As his talks with employees changed, Bechtel saw that the idea of a virtual assistant was becoming popular. “At first, they would say, ‘Joe, you’re insane, you cannot automate those processes,'” he recalls. But then people began asking, “What should I call my bot?”


Although it might seem insignificant, Bechtel claims that naming the bot was a crucial step in the adoption process. We needed to carefully brand our automation so that people wouldn’t mistake it for a passing trend and instead see it as a collaborator who would change with them. These conversations gave rise to the name “Allie.”

Allie is a smart coworker who will not stand still or do nothing. “Allie will develop and mature alongside the colleagues as they master new skills,” Bechtel said. So, when their tasks change and get more complex, they should not be using a swivel chair or copying and pasting. As our software evolves, Allie will also develop and gain new skills as we layer in document processing and more [natural language processing] tasks. We took a really intentional approach to doing it.

As a true employee, Allie goes well beyond simply crunching numbers or producing reports. “We start off with design thinking, like how we train both users and the system,” Bechtel explains. “We would hold art-of-the-possible sessions and teach them the best flows and use cases for automation.”

For instance, Allie currently supports clinicians in the field of claims. Bechtel explains, “In the past, you would utilize medication and therapy off-label. “To approve or deny the claim, the pharmacist in this situation would need to research the data and exercise sound judgment. They lacked access to data on the historical approval rates of that claim based on the patient’s demographics. Allie may now go through and compile that data, including the acceptance or rejection percentage for drug coverage.

“To aid physicians, not to remove them from making a judgment, but to complement them,” he continues, “so they may have all essential information at their disposal.”


Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.