Those who visited our stand at the Singapore Fintech Festival this month would have seen a demonstration of Privé’s latest AI initiative, the Portfolio IQ. This is the latest iteration in Privé’s mission to bring personalisation to complex portfolio processes in wealth management.
Wealth managers today must cope with information overload, interpret and follow market views all the while ensuring that investments are suitable for their clients. They must also be able to explain the rationale behind investment decisions or recommendations. On the other hand, regulators are increasingly cautious around the output of AI models and veracity of generated text. Privé is therefore putting significant early efforts in tackling these challenges.
Currently a proof of concept, Portfolio IQ builds on the tried and tested AI GO (Genetic Optimiser), Privé’s patented digital portfolio construction tool. Portfolios are optimised using genetic algorithm concepts, according to various ‘Fitness Factors’ or constraints. Privé has already successfully implemented AI GO with a number of banks and insurance companies since 2017. The tool assists advisers in constructing portfolios for clients utilising model/target allocations (‘house view’) and a client's asset recommendation list.
With Portfolio IQ, this concept is taken a stage further, using large language models (LLM) to interpret directional inputs in the form of text - whether structured or not. In the first stage Portfolio IQ takes a quarterly CIO Outlook, infers preferences for particular asset classes and sub asset classes, and translates this into vectors. These are then used to determine the initial weights of different assets, before other constraints are used to optimise the final portfolio .
At the second stage, a complete proposal is then generated for the client’s portfolio, with explicative text and talking points, enabling the Portfolio Manager, or Advisor to link the insights in the Outlook report with the proposed investments. These outputs may be configured to implement changes in an existing portfolio, invest cash injections into a new portfolio, or propose trade recommendations for Advisory.
Following this successful proof of concept, the technology may support various other use cases on which Privé is currently working, with the ability to enhance existing modules in the Privé Professional workspace for portfolio managers, sales staff and developers such as:
- Investment Explorer
- Market Commentary
- Portfolio Reporting and Gap Analysis (Alerts)
- Answering Internal and External Queries (Chatbots)
- Meeting Preparation and RFP Drafting (Sales, RMs)
- Product Functionality and Documentation (Developers)
Watch this space as Privé continues to explore how LLM may be used to supercharge wealth managers in their daily tasks, while keeping abreast of regulatory developments in this area.
 These may relate to asset class or geographical limits, currency, sustainability criteria as well as measures to reduce transaction fees or ‘churn’ in a portfolio by re-using existing assets.
Disclaimer: This press release is for information purposes only. Some of the statements in this press release may be forward-looking statements or statements of future expectations based on currently available information. Such statements are naturally subject to risks and uncertainties. Factors such as the development of general economic conditions, future market conditions, unusual catastrophic loss events, changes in the capital markets and other circumstances may cause the actual events or results to be materially different from those anticipated by such statements. Privé does not make any representation or warranty, express or implied, as to the accuracy, completeness or updated status of such statements. Therefore, in no case whatsoever will Privé and its affiliate companies be liable to anyone for any decision made or action taken in conjunction with the information and/or statements in this press release or for any related damages. Privé assumes no obligation to update any information contained herein.