The use of artificial intelligence in fund administration is all about being smarter with data, with many organisations now trying to transition their analytics from being descriptive to forward-looking, according to Jay Cipriano, senior vice-president at SEI Investment Manager Services.
“One of the attractive things about AI is its ability to power predictive analytics. Predictive insights enabled by machine learning could prove to be transformative for many firms, leading to better decisions across a range of functions,” he says.
General partners are looking to combine their asset and investor data with market data to predict future opportunities – it can inform them about what products or funds to offer with expectations of high market demand, anticipate flows that affect their investors and funds, and help them prospect more efficiently, Cipriano adds.
“Having data doesn’t necessarily make
MUFG Investor Services
Over time, it will become easier to take advantage of the data that’s available and get the most information out of it. But currently, one of the challenges is gaining access to enough data to develop AI or machine learning.
“By design most data is kept private within a very small circle of fund managers and investors. To date, it has been very hard to accumulate large amounts of data which one could analyze or try to use in some sort of quantitative investing strategy,” says Cipriano. “As transparency increases and private markets become less private, it is easy to see how AI can become more critical. New sources of structured and unstructured data crop up with some regularity, and combined with machine-learning technology, AI offers the promise of insights that were unimaginable only a few years ago.”
Natural language processing
One of the main focuses for fund administrators has been natural language processing, which can be used to extract data from the myriad documents that are needed to run a private equity fund, such as financial statements or subscription files.
“These are big documents and we don’t have control over how we receive them, so for us to be able to seamlessly process that information is a big win,” says Mike Dickey, global head of product development at MUFG Investor Services.
So far, Dickey says, the work the firm has done to automate processes like capital calls or distributions of capital statements, has allowed it to cut its operating costs in half, but the next goal is to bring everything closer to how hedge funds operate.
“I think everyone in our industry is on a bit of a journey with data, some firms are further ahead than others. But having data doesn’t necessarily make it useful. Unless I can normalise that data, I can’t make use of it,” says Dickey. And once the data are easily available and ready to use, then firms can use machine learning and AI to forecast trends, for example to find out what investors might be at risk of leaving a fund.
However, it’s still early days, according to Alex Di Santo, group head of private equity at Crestbridge. AI is being used to scrape data from financial documents, to monitor investor behavior on platforms where the GPs can see what information LPs are reviewing and assessing and in deal sourcing.
Increasingly, clients are demanding more, though, and Di Santo sees a future where fund administrators will have all the processes and straight-through-processing tools together in one system where AI is used to link it all together.