VIDEO | Often said to be one of the sectors most resistant to digital transformation, private markets have a lot of catching up to do – and its players are recognising the fact. At the PostTrade 360° Nordic 2024 conference, the session titled “The operational gains from AI, data, and cloud use cases in a private assets post-trade ecosystem” took a deep look at how technology can make a difference to sector.

The session opened lightheartedly with a joke from Matthew Goldblatt: the private market is the “ultimate post-trade asset class” – not in the meaning of post-trade we are familiar with, but because transactions were made over the post.

The head of investment operations solutions sales EMEA and APAC at JP Morgan recalled that there used to be a mailroom in his office which sent valuation statements to investors “45 to 90 days after the quarter has ended for those assets”. This meant that even investors with large investments in companies, properties, and venture capital “who have committed tens, if not hundreds of millions of dollars to these assets” were receiving notices only four times a year over the post.

Advertisement

Watch the session video here!

When logged in, pick up your free access pass for this Nordic conference if you haven’t. Then browse back to this page, and the video will show up in this box.
(Don’t have an account here yet? Register here. This is separate from the event platforms where you may have been signed up as a delegate.)
Login

AI is here

Change is coming, however. Goldblatt observed that the deployment of artificial intelligence (AI) – in particular, natural language processing (NLP) – into private market operations has stepped up in the last two to four years. He believes that for the teams responsible for processing statements and data, being able to use AI to retrieve necessary information would be “very much an operational game changer”.

Rob Calder, enterprise sales executive at Canoe Intelligence sees potential in AI for “scalability, speed, and accuracy”. Citing the flagging of exceptions as an example, he pointed out that once an AI model achieves 95 per cent straight-through processing, users are simply dealing with the exceptions rather than the processing of everything.

Bigger, better

Before JP Morgan worked with Canoe Intelligence to leverage AI in its operations, a team would have to manually pull data from documents on the behalf of fund managers and the bank’s institutional clients. Today, however, the work of the initial retrieval and data scraping is handled by NLP. “It’s an augmentation of the people who have had historically done the work,” said Goldblatt.

“It is the ability to not only be a touch more efficient and accurate, but most importantly, scalable, so you can add volume,” he continued. “We are able to support the clients’ month-end reporting in a more robust fashion and pull more timelines in.”

The ability to handle large volumes of data is an important advantage of using AI. Goldblatt illustrated, “Instead of having a person working eight hours a day processing a particular pension fund’s 70 PCAPs (partners’ capital account statements) for the month-end, there is no human limit in terms of the number that it (AI) can take.” JP Morgan has set up the AI it uses to scrape data three times a day, going through document portals and pushing through relevant information “in a matter of seconds”.

In addition, the data has also become richer. The AI can be tasked to process many more data fields compared to a human – “We wouldn’t ask an analyst to record everything even though we only need three fields,” Goldblatt pointed out – so the data in the statements today can go beyond “the capital being called or the valuation being reported”.

No quick success

Despite AI’s efficiency, we are far from fully automated banking. Calder estimated that AI might “save 80 per cent of the headcount” in the core function of collecting documents and extracting data. Goldblatt countered that, saying that in practical application at JP Morgan, it was closer to “a 50 per cent reduction in the use of staff”.

An operations team is still required to oversee the AI. “It does require a teacher, or a guardian”, he pointed out. He shared that in the initial beta client testing, the JP Morgan’s AI programme achieved only 23 per cent accuracy. The figure gradually increased to 60 per cent over months. Likening the process of training the AI as putting it through university, he reminded the audience that training the tool to get to the industry standard of 95 per cent accuracy takes time and patience.

Panellists:
Matthew Goldblatt, Head of Investment Operations Solutions Sales EMEA and APAC, JP Morgan
Rob Calder, Enterprise Sales Executive, Canoe Intelligence

Moderator:
Lidia Jast, Head of Sales, Securities Services, Sweden, JP Morgan

Hosted by JP Morgan.


• The consolidated PostTrade 360° Nordic conference, in Stockholm on 4–5 September 2024, came to host 1,200+ delegates and featured 70 sessions.
• Log in and watch the sessions – for free! Links in the grey video boxes on each article will guide you intuitively. Yet, if you still want watch the 2-minute intro to our video offer, click here.
• Our coverage relating to the event is indexed here.
• By the way … are we connected on LinkedIn already? Follow us here.