February 28, 2019 | Written by: Peter Ivankay
Categorized: Customer Stories | Watson
Share this post:
Digital assistants and the AI sweet-spot
It’s a well-known scenario: a difficult question comes up in a client meeting and the one person with the expertise to answer isn’t there. But what if you could call on that person’s expertise anywhere, at any time, and gain access to their trusted knowledge and insight via a digital avatar?
UBS recognizes the potential to modernize our approach to client relationships as technology and expectations evolve. Looking for the “AI sweet-spot,” we realized that humanizing digital assistants would help us make that transformation, and worked with FaceMe and the IBM Cloud team to create digital avatars.
Redesigning customer experiences
IBM and FaceMe helped us develop two digital assistants, used exclusively during face-to-face meetings in one of our UBS offices in Zurich to help advisors and clients of our wealth management business. The first assistant, Fin, is a friendly character designed to handle simple procedural queries—for example, helping a client or their advisor request a replacement for a lost credit card. By making these relatively routine tasks faster and easier, Fin helps keep advisors free to focus on the higher-value business of optimizing each customer’s investment portfolio.
The second avatar, named Daniel Kalt, is a digital replica of our own Regional Chief Investment Officer for Switzerland. Daniel’s avatar is designed to understand and answer questions about markets and the current outlook for investors, drawing on a trove of historical data and on the bank’s latest research. In representing a person whose knowledge is trusted and respected by both UBS employees and our clients, the Daniel Kalt avatar helps clients feel more informed.
The avatars draw on existing data and procedural routines, augmented by the UBS “house view”—that is, the bank’s own outlook—of opportunities and risks in wealth management. This house view is updated regularly to reflect ongoing industry developments and market trends and is one of the primary sources of investment advice we offer to our clients.
Interacting with Fin or Daniel
UBS uses the FaceMe platform to allow interactions with Fin or Daniel through the use of voice commands. The avatars appear on a TV screen in the client meeting room, powered by the FaceMe platform’s distinguished, real-time animation capabilities.
FaceMe’s platform, hosted by IBM Cloud, uses IBM Watson Natural Language Understanding to process questions asked in natural language and determine the intent behind them. Using situational awareness together with stored expert data from IBM Watson, the system creates an appropriate response which is delivered as synthesized speech by the FaceMe-created digital avatar.
Initial responses to Fin and Daniel have been positive; many of our advisors report that having a digital version of Daniel Kalt on speed-dial is a real “wow factor” for customers. The user experience is smooth and intuitive. Adding to its speaking capabilities, digital Daniel Kalt can also output various graphs and other visual aids to help users better process complex information streams.
FaceMe’s platform allows the deployment of digital assistants at scale over the web, where avatars are rendered remotely, in real time, on IBM Cloud servers with GPUs.
Rise of the cobots
Our UBS Companion project is in the prototype phase. Once we define the most powerful use cases, we believe Fin and Daniel have enormous potential as “cobots” to help facilitate frictionless access to UBS expertise and to enhance the speed and quality of customer service interactions.
The end-to-end response takes just 200 milliseconds or less, from capturing the audio, to processing the question and articulating the response. Fin and Daniel can respond to certain questions as quickly as their human counterparts, with perfect recall and consistency, reinforcing the service provided by human advisors and demonstrating to clients that UBS is maintaining technology leadership.
We look forward to continued support and development from both IBM and FaceMe as we explore this exciting new technology.