READ NOW: Singletrack Quarterly Benchmark Report Q1 2024

The role of AI in capital markets advisory: summer client roundtable


Last month, Singletrack met with representatives from a handful of our clients in both London and New York to discuss the role of AI in capital markets advisory. Our regular client roundtable discussions are one of many initiatives we run at Singletrack to provide a forum for our people to discuss key topics facing the industry.

It’s clear that AI will revolutionise capital markets and it’s developing extremely quickly. Discussion in both London and New York was lively and engaging. Here are the key takeaways.

  • Sell Side Firms are acting now
  • Big questions include availability of high quality training data, compliance, use case identification and workflow integration
  • AI will structurally change capital markets over the next three years
  • Find out how you can harness AI tools for capital markets now

Firms are acting now

There has been some news already about advisory applications for AI tools. Many we spoke to have personally been experimenting with tools such as ChatGPT, and some firms have granted access to teams in order to facilitate this. Everyone was open to incorporating AI into workflows, and felt that doing so was probably inevitable. 

Compliance is a key issue, and there is a lot for lawmakers and regulators to catch up on. The risks and responsibilities associated with using generative AI are still unclear and content and data ingestion to train models was a big question for our attendees.

For example, can we be sure that the open-source LLMs currently available, some of which are specific to capital markets, will really ensure data stays within a particular firm’s environment? And where firms restrict themselves to training generative AI on proprietary content, is there a risk that this narrows the output, proliferating mistakes or so-called hallucinations?

This question is only more pressing when we take into account the specific needs of capital markets firms: will LLMs emerge with a sufficient level of specificity to meet them? Another current difficulty is the current lack of commercial APIs.

And finally, what is the effect of generative AI training on the value of research content itself? Will it make research content a more sensitive resource, since an insecure library might be used to generate further content based on a firm’s output?

AI use cases

There was a lot of interest in AI’s potential as a time-saving tool for both internal and external tasks such as summarisation, research and automated communication in the future. But there’s a fear that, for the moment, firms can’t realise the time savings AI promises because there will still be a need for a trusted person to check output.

In some cases, like automating certain client communication, this may not be a problem: having someone sanity-check emails before they are sent may still represent a net efficiency compared to manually creating them. But for tasks like research or summarisation, checking the AI output could be even more time consuming than just doing it manually in the first place.

For the moment, most are looking at using generative AI for internal rather than external tasks, and are more focused on analysis rather than generating output. In addition, firms are tending to lean on their partnerships with technology solutions vendors to identify and integrate AI into the tools they’re already familiar with.

Will the big initial investment be worth it? And when is the perfect time to make such an investment?

Efficiency or revenue growth?

There was a consensus that firms are looking to see efficiency gains from AI at the moment, but expect a shift to revenue growth over time as solutions mature and industry familiarity develops.

An example of where many would like to see AI assistance is in interaction logging. For many firms there is a tension between the highly valuable data generated by detailed logging and the time it takes to perform it. 

Panellists agreed that tools which reduce the friction will lead to more comprehensive, higher quality data, and in turn, to better results. This is an area where AI-backed tools would be very welcome.

What will the structural effects of AI in capital markets be?

AI tools are certain to be transformative for capital markets, as for many other industries. There are big opportunities, many of which may not even be visible yet, to gain an edge by adopting AI-backed tools. 

Singletrack incorporates a range of AI-backed tools into our products right now, and we are working hard to ensure our clients can benefit from cutting-edge tools as they become available.

To find out how we can help your firm harness the power of these tools, get in touch now.



Published: 19/07/2023