Artificial Intelligence

UBS is testing a cognitive agent on its back office employees

  • UBS is testing the use of a virtual human to complete complex tasks quicker and more efficiently
  • UBS is working with software company IPsoft to use a cognitive agent called Amelia, a tool it hopes can help support employees by next year
close

Email a Friend

UBS is testing a cognitive agent on its back office employees

UBS is testing an intelligent virtual assistant on some of its back-office employees, with the hopes that it will eventually help customers navigating through the UBS website.

The bank wouldn’t say when Amelia, the virtual employee’s name, will be deployed to help customers. But UBS said it hopes the virtual employee will help customers retrieve lost account information when they can’t log in and set up bill payees. Meanwhile, a pilot with employees has so far yielded encouraging results.

“We did not share that they were interacting with a cognitive assistant,” said Tom DeCarlo, managing director and head of client services at UBS. “[Many] thought they were talking to agents on the floor.”

UBS began an initial U.S. pilot with a dozen client sales assistants in mid-January of this year, an exercise which wrapped up in March. The bank said it plans to get the application up and running for its client sales assistants around the world by 2018. An additional Amelia pilot with the bank’s internal tech desk is currently underway in the U.K. and another one is planned for Switzerland. UBS working with software company IPsoft, which developed Amelia to help finance companies more efficiently deliver service to clients.  For the UBS U.S. pilot, DeCarlo said Amelia worked with client sales assistants responsible for moving funds from deceased account holders to beneficiaries. Amelia helped UBS staff confirm information and fill out forms.

“The folks that take those calls within the service organization today have roughly 10 to 15 years of experience, at the desk or within operations as there are many different variables that could come into play when outlining the document requirements and signatures relative to deceased retirement distributions,” DeCarlo said.

The move is part of a larger push from UBS on automated service enhancements. For instance, last year, the bank announced a pilot program with Amazon Echo to let clients and non-clients of the bank get answers to financial and economic questions. And the bank has been using basic bots to handle simple back-office tasks for over a year.

While Amelia has voice capability, during the pilot, UBS employees communicated with Amelia through a chatbot interface. Compared to basic bots, Amelia has the capability to deal with unstructured data that requires thinking and organization, which could be a major advantage in helping the institution deal with complicated processes and documents. And Amelia’s ability to think on her feet lets her to pivot easily based on learned behavior. “There’s two types of learning, that which is observed from rules, and there is also event-based episodic learning — if you have episodic learning, agents can learn very adeptly,” said IPsoft CEO Chetan Dube in a recent interview with Tearsheet. UBS’ emphasis on virtual human assistance dovetails on efforts other major banks have been taking, including Bank of America’s virtual assistant Erica.

The bank said bringing Amelia on board won’t result in any job losses, as it sees people migrating to new roles as they work alongside the technology. And since Amelia is fluent in a dozen languages, the bank can easily deploy her to help employees and customers around the world. DeCarlo said those who fear getting stuck when Amelia can’t resolve an issue can take comfort in her ability to sense when to escalate matters to humans.

“She’ll ask additional probing questions until she gets to a point where her confidence level is less than 99.9 percent, and then she will conference in a supervisor or subject-matter expert.”

 

0 comments on “UBS is testing a cognitive agent on its back office employees”

Artificial Intelligence

A deeper look into what makes a successful chatbot with Bank of America’s Erica

  • Erica reported 19.5 million users, over 100 million interactions and 90% efficacy for useful answers.
  • BofA’s Hari Gopalkrishnan says that the key to Erica’s success is a balance of personality and functionality.
Minahil Shahab | August 26, 2021
Artificial Intelligence, Member Exclusive

Financial anxiety is clogging users’ financial activities — could automation be a solution?

  • A significant portion of U.S. consumers struggle with managing their finances because of financial anxiety.
  • With AI becoming increasingly advanced, autonomous finance could take some of the load off.
Rivka Abramson | May 26, 2021
Artificial Intelligence, Member Exclusive

Human financial experts may be too challenging to automate — hybrid solutions are the next-best thing

  • For now, humans continue to offer financial expertise that can’t be automated.
  • So instead of trying to replace financial advisors altogether, companies are focusing on automating the gap between users and experts.
Rivka Abramson | February 02, 2021
Artificial Intelligence, Member Exclusive

Far from changing banking, chatbots aren’t keeping up with call volumes in the COVID-19 era

  • More financial institutions are using chatbots to handle increased customer service loads precipitated by coronavirus.
  • But, are those chatbots really providing meaningful experiences and value?
Sara Toth Stub | June 29, 2020
Artificial Intelligence, Podcasts

Flybits’ Hossein Rahnama: ’90 percent of digital transformation budgets are spent on managing IT complexity’

  • Banks struggle to provide automated recommendations.
  • Flybits wants to remove the IT complexity that eats away at their time and budgets.
Zack Miller | August 27, 2019
More Articles