UBS is a Swiss multinational investment banking and financial services company ranked 30th on S&P Global’s list of the top 100 banks. In addition to investment banking and wealth management, the company is looking to improve its tech stack through several AI projects.
Our AI Opportunity Landscape research in financial services uncovered the following three AI initiatives at UBS:
- Virtual Financial Assistant with Digital Avatar: UBS’ virtual assistant project developed by a third party.
- ORCA ML for Foreign Liquidity Gaps: An internally developed ML platform for automatically solving gaps in foreign exchange liquidity.
- Cognitive Search for Wealth Management and Investment Banking: AI-enabled enterprise search using natural language processing (NLP) customized for financial services and investment banking.
We begin our coverage of UBS’ AI initiatives with their project for a virtual financial assistant for their banking clients.
Virtual Financial Assistant with Digital Avatar
UBS partnered with IBM and Digital Humans (formerly FaceMe) to create a virtual financial assistant for its customers. The virtual assistant is a conversational interface built with IBM’s Watson Natural Language Understanding solution. Watson runs primarily on natural language processing technology, which is an approach to AI that enables the extraction and analysis of written text and human speech. Digital Humans provided the 3D character model for the avatar, which represents the assistant on-screen.
The video below explains how Watson Natural Language Understanding works:
UBS developed two distinct digital avatars. One avatar, named Fin, is built for managing simple tasks such as helping a customer cancel and replace a credit card. The second avatar, Daniel, can purportedly answer investment questions. IBM claims Watson affords UBS the following capabilities:
- The ability for financial advisers to ask questions about markets and if the time is right to invest. IBM claims its Watson software answers these queries based on historical market data.
- The ability to customers to ask the virtual assistant to request a new card for them, and the assistant will automatically send the request and cancel the customer’s current card if need be.
- The ability to save time on the most common customer service tasks such as updating identification information. Additionally, customers could ask to be directed to the correct online channel to complete their task such as a company website or mobile app.
ORCA ML for Foreign Liquidity Gaps
UBS also started an internal initiative with the goal of solving liquidity issues within foreign exchange using machine learning. In 2018, the bank announced its ORCA direct solution, which purportedly helped its employees execute foreign exchange transactions more quickly.
The bank’s software could automatically decide the best digital channel by which to execute a foreign exchange deal. This may save the bank a significant amount of time, as it would be particularly difficult to optimize for a bank with access to so many separate trading channels.
Additionally, these platforms may run on different pricing metrics, and banks may incur certain fees depending on the type of trade they are making. UBS updated the solution to ORCA Pro in 2019, which it claims can now act as a single-dealer platform.
This platform is linked to UBS’ optimization engine, which helps reduce disparity between the expected price and the price at which a trade is executed. For example, if a given deal is made weeks after UBS’ financial advisor had last spoken to the client, ORCA pro might be able to discern that the bid/ask spread for the deal has fluctuated without either party noticing.
UBS claims their ORCA Direct and Pro solutions provide the following capabilities to their staff:
- Price estimates for each transaction. These estimations are based on a calculation of liquidity across multiple channels and how long that liquidity might last.
- Trade channel optimization. UBS claims their software can optimize a trade according to price and which channel can process the payment most quickly.
- Trade visibility for clients. UBS’ FX clients can now purportedly access the same amount of trade information as their brokers, enabling quicker and more informed trades.
Cognitive Search for Wealth Management and Investment Banking
UBS’ third AI initiative is their partnership with vendor Attivio to develop an NLP-enabled search engine for their wealth management, asset management, and investment banking services. Attivio refers to this NLP-based solution as cognitive search, which can be understood as an AI-powered enterprise search application.
The short, 1-minute video below explains how machine learning can enable enterprise search and provide context for more detailed results:
The vendor claims UBS developed this application to facilitate the following capabilities:
- Unified enterprise data, accessible by all knowledge workers so they don’t need to search in multiple locations for documents.
- Proactive insights based on the results surfaced from the cognitive search solution. For example, a UBS employee might find a relatively sudden spike in trades to entities in Japan while trying to optimize a single deal with a Japanese client. This might give the the employee context into how UBS is currently working with their Japanese clientele.
Emerj for Financial Services Leaders
Financial services companies need to understand what their competitors are doing with AI if they hope to compete in the same domains and win the customers their competitors are trying to court with more convenient experiences and more financial lucrative wealth management services.
Leaders at large financial services companies use Emerj AI Opportunity Landscapes to discover where AI can bring powerful ROI in areas like wealth and asset management, customer service, fraud detection, and more, so they can win market share well into the future. Learn more about Emerj Research Services.
Header Image Credit: UBS