When deploying generative artificial intelligence (Gen AI) technologies in the financial sector, banks and financial services firms are primarily concerned about data privacy which means making sure that they do not unintentionally surrender proprietary information, as well as the perceived risk that when using this technology, information could be transferred to unknown third parties.
Another concern of potential financial industry users is that, as they need quick answers to complex questions, it is important for them that the responses they get from Gen AI are accurate enough for them to confidently make business decisions.
Finally, there are concerns that the regulatory landscape around Gen AI is still trying to catch up with the technology.
And while many people may have dipped their toes into the technology purely for entertainment value – for example, via the recent release of Generative Pre-trained Transformers, commonly known as GPT, via ChatGPT – it is generally accepted that the technology, when correctly focused and purposed, will open up new levels of efficiency and reinvent traditional applications.
Speaking to The Asset at the recent Singapore Fintech Festival, Joseph Lo, head of enterprise platforms at global fintech firm Broadridge, points out that GPT technology has already begun to revolutionize financial services.
Lo highlights Broadridge’s BondGPT, a tool that helps bond traders uncover information about bond options in real time, condensing work that used to take 20 minutes into two minutes.
Since BondGPT’s introduction in June this year Lo says the Gen AI-backed platform now has more than 700 users at over 200 firms globally. “Many clients have begun to realize how powerful BondGPT could be, especially if it incorporates their own data in their existing systems and workflows, and we are in talks with several top-tier asset managers about how BondGPT can play an even larger role in supporting their users.”
With rising interest rates, clients have been seeking better intelligence and pre-trade analytics in the corporate bond market, and Lo notes that strong demand for a tool that brings all this information together was met by Broadridge’s Gen AI solution.
“This is why BondGPT was launched,” he states. “BondGPT is the first application in the market powered by OpenAI GPT-4 that answers bond-related questions and assists users in their identification of corporate bonds on the [Broadridge’s] LTX electronic trading platform.
“It uses the LLM [large language model] chat function that allows users to ask questions and identify corporate bonds on the LTX platform based upon the user’s criteria. If a user wants to know the price of a bond, or what is happening in the market to make the best decisions for their customers, BondGPT is able to retrieve this information in real time from many different data sources at once.”
The ability to receive quick responses to complex bond-related questions, Lo adds, helps market participants to inform and expedite vital pricing decisions, facilitate counterparty selection and broaden liquidity access.
AI monster?
But in creating BondGPT, Lo doesn’t think he may have created an AI monster that will quickly replace traders after downloading their decades of knowledge and habits. In fact, he notes that his firm in September fearlessly doubled down on the technology.
“We made GPT-4 available for all Broadridge staff, and everyone has a chat agent to assist them in their day-to-day work,” he points out. “And we are also investing in the technology to help our software engineers increase efficiency so that they can also provide more effective solutions to our customers at a faster rate.”
With this strategy in place, Lo says, his team can continue to explore the practical application of AI, whether it’s to facilitate trading across other asset classes beyond bonds or providing different data for other products and services.
The creation of BondGPT, he adds, has allowed his company to really see what users need and want in their trading process. “A key piece of feedback we have received from our clients is that they want to incorporate their own data. And, in addition to the advanced analytics already on the platform, they want more control and personalization.
“We quickly took on this client feedback and delivered BondGPT+, the enterprise version of BondGPT, only four months later in October. This not only meets our clients’ enterprise requirements, but also sets a new standard for generative AI solutions in the financial industry.”