Bloomberg and TP ICAP set out some use cases
Image:
Financial services companies are experimenting with generative AI
Much has been made of the potential of generative AI in 2025 and there’s a great deal of money riding on whether this technology delivers. However, amidst all the noise, some companies have been quietly experimenting, with emphasis on solving real-life problems.
Since Chat-GPT arrived towards the end of 2022, it’s been rare to have a technology focused conversation which doesn’t take in the potential of generative AI. But for many businesses, potential is where they’re stuck. The use cases have been slower to materialise than many hoped – particularly the investors who were promised such wild returns.
Dr. Amanda Stent is Head of AI Strategy & Research in the Office of the CTO at Bloomberg. Speaking with Computing earlier this week, Amanda said:
“This year in the US, 41% of venture capital money has gone into generative AI and that is an insane amount. As to whether the potential matches that investment, well that is literally one of 2025’s biggest challenges. The technology has to make money, and it has to be really useful for people.”
When viewed from this perspective, generative AI doesn’t seem to have quite shaken off its “solution in search of a problem” label, which is a problem when in many businesses, budget owners are under pressure to pull off a digital version of Jesus’s loaves and fishes trick.
Image
Description
Dr. Amanda Stent, Bloomberg
“ We have more than 300 AI experts at the company,” says Stent. ”I would call all of those people hammers looking for nails. It’s a question of finding the right nails, and the right nails are our clients’ real problems.
“Our approach is to figure out what our clients’ needs are and then figure out how we can address those needs using AI, if necessary. AI introduces complexities that traditional algorithmic methods do not introduce such as a lack of determinism and in the case of generative AI, the hallucination problem.”
That said, as 2024 has progressed, it’s become clear that businesses are quietly experimenting with LLMs and fine-tuning those models with their own data. Mindful of the old but evergreen GIGO adage, they may also have been focusing on their underlying data architecture to ensure that data is clean, accessible and in a usable format.
Indeed, John Kain, Head of Financial Services Market Development at AWS, speaking to Computing at AWS re:Invent last week said:
“The customers that have been most successful at improving the user experience are those that made be early investments in enterprise data architecture on the cloud, that really unlock the underlying data to allow them to do what they want on the user experience side.”
Bloomberg made AI-Powered Earnings Call Summaries available on Bloomberg Terminal® earlier this year. This uses the pragmatic application of artificial intelligence (AI) to help analysts with their research process. Amanda Stent explains why it was necessary:
“Earnings calls are very long, and to discover the piece of information that you might be interested in could take a really long time. We have a product that does thematic summaries. Our subject matter experts identified a set of themes, and we construct bullet point summaries on each theme.
“I could construct a summary of an earnings call but, as a technologist, the one that I would produce would be fairly generic. With the involvement of our subject matter experts such as our product teams and our data teams, our sales force, we can create products that are really targeted at financial services, and a product that’s really differentiated.”
People who understand the technology don’t understand user problems
Somone else with a clear focus on the problem-solving potential of generative AI is Max Spoto, Group Chef Operating Officer at liquidity and data specialist TP ICAP. Speaking exclusively to Computing at re:Invent about the company’s recent collaboration with AWS, Spoto said:
“It’s very clear that this type of technology has an enormous amount of potential but it’s also clear that the people who understand the technology do not understand the problems that users are trying to solve – and vice versa.”
Image
Description
Max Spoto, TP ICAP
“It was easier for me to convince a technology person to follow me and my clients in my P & L because as technology people they’re generally curious. It’s more difficult to get a businessperson to understand the technology. Some of what we’re trying to do with this partnership is affect a shift. So, we’ve been working with corporate functions first not trading businesses because when you deal with trading businesses you have compliance, risk and other issues.
“That’s why we’ve confined these AI experiments to the corporate function – operations, technology, finance, HR, legal. We’ve had a fair amount of success such as metadata PDFs for procurement and a chatbot on Salesforce to gather customer insight.”
In addition to the automation of regulatory governance reviews, Spoto was able to share some other use cases.
“We are using this technology to deploy code faster and to convert unstructured date into structured. That on any given trading and dealing floor is quite an opportunity because there is a lot of voice, chat and other unstructured data. We think that will over time give us a lot more insight into pre-trade. There’s a lot of voice-to-text type technology as well and potential for digital assistants or agents for things like trade summaries, trade recaps – all trying to alleviate trading workforce pain.”
TP ICAP is establishing an AI and Innovation Lab with AWS to accelerate and scale AI-driven solutions. In the longer term, Spoto foresees the individual use cases outlined above coming together with others into a much smarter operation with a far more cohesive and flexible data fabric underneath.
“We are in the business of market data liquidity and so we deal a lot with market data, client feedback, news events, corporate events. Over time as the technology becomes better and faster getting all that data into one place and able to accurately curate customers insights and make our people smarter, faster.
“I think that what is interesting is that this type of technology not only solves today’s problems but gives you a feel for future solutions that you might not have even thought of yet. But you have to experiment slowly, and our plan is to make sure we do this from a user’s point of view.”
The first article in this series, discussing how generative AI is enabling old, established financial institutions to compete with fintech disruptors can be found here.
