Private Debt Investor

Private debt still slow to embrace AI

By Claire Coe Smith

November 1, 2024

‍

If private credit firms think they should be doing more to embrace artificial intelligence, the truth seems to be that even most of those that are already on board are so far only scratching the surface. Adoption remains pretty low across the asset class and the funds taking first steps are typically starting with tools for portfolio management and reporting rather than integrating emerging tech to drive better investment processes.

Eric Burl is head of discretionary at Man Group and was involved in the fund manager’s recent acquisition of US mid-market private credit manager Varagon Capital Partners. Already, Man has hired three tech people dedicated solely to bringing tech, quantitative computing and AI tools into that business.

“Having met with a lot of private credit firms, public and private, over the last few years, it is interesting to see how few of them use any technology at all, let alone in investment processes,” says Burl. “This is in Man Group’s DNA, so we see lots of opportunities to enhance processes when we look at those managers – many of them are ripe for improvement in relation to technology, AI and quant.

“The lines between quantitative systematic investment and discretionary investing are blurring. We don’t see computers replacing humans but they can massively enhance investment processes from both an efficiency and an accuracy perspective.”

Heavy lifting

Man Group has already built a natural language processing model to read the PDFs that Man Varagon analysts receive from borrowers each month and to aggregate the data into a single report for the portfolio. “We have built a model to go in and read those PDFs, reap the data we want and produce that report,reducing the time analysts spend on that to a couple of minutes,” says Burl.

In the future, AI could be doing a huge proportion of the heavy lifting across the investment process. At Liquidity Group, an AI[1]driven asset manager backed by Apollo that focuses on growth credit to technology businesses, the firm is using patented technology to analyse technology companies with more speed and certainty than its peers.

The platform can scrutinise company financials, assess performance and predict future outcomes at potential borrowers in minutes, allowing Liquidity to make snappy, transparent and unbiased lending decisions and issue term sheets in as little as a few days.

Liquidity has $2.5 billion under management, and committed $1.5 billion to more than 50 companies last year alone. Carmen James, chief strategy officer for the business, says: “Undisciplined decision-making at scale erodes the quality of that decision making over time. We have built out a growth credit underwriting model that allows us to plug in directly to a company’s financial data and analyse all the key metrics that any other credit provider would look at. We can understand exactly where that company sits versus its peers, the risks it faces and how we should structure our facility; the tech gets us from zero to day 60 of an investment process in two clicks.

So far, Liquidity claims to have delivered a zero percent loss ratio for LPs over the past three years, and by relying on technology to underwrite a segment of the private credit market that is viewed as one of the most risky, the firm hopes to prove the technology for more use cases.

The majority of Liquidity’s 150 staff are in research and development or data science, with a lean origination team of about 15 people leveraging technology to build relationships and personal networks. “We take an extremely targeted approach to relationship building,” says James. “We use technology to analyse the firms we want to target and to make sure we are connecting with the most relevant people. It is not man versus machine; it’s man plus machine.”

Data interrogation

Aidan Kenny, head of innovation at law firm Maples Group, says most debt funds are just scratching the surface: “One exciting area is AI’s ability to analyse vast data sets from multiple sources and interrogate that data,” he says.

“That allows you to very quickly monitor and analyse a lot of complex information to make faster decisions in areas such as assessing potential loans, for example. Another area is data extraction, where you have a lot of documents and can use generative AI to intuitively extract information without the need to do that manually.”

Kenny says AI will be more important on the creative side of things, as when testing a hypothesis or a scenario, AI can add to the conversation by creating scenarios that might impact a certain position or process. This can be used to test hypotheses and to digitally prototype ideas before implementing them, he adds.

Man Group’s Burl argues that the real challenge, particularly for the more traditional private credit managers, is data. Access to data and availability of data is nowhere near as granular or as frequently delivered as it is in public markets. “The data needs to catch up with the quality of the technology to allow us to do the work we want to do,” he says.

Building diversification into portfolios is another area where AI could shine, given its ability to analyse lots of companies and deliver a lot of data on those companies. Moving forward, private credit will need to lean on data providers to get that kind of information in a more standardised format to achieve that goal.

Burl gives a further example: “One thing we do on the public equities side that is hugely valuable is taking a manager’s trading data and using that to analyse what they are good at, what they can improve and what they should stop doing,” he says.

“That is easier in equity because companies only have one stock, but we have been building that on the debt side to cope with the added complexity of the capital stack. We are building that in public credit and it would be amazing to extend that to private credit. There is not as much trading and not as many data points, but we should still be able to capture what we can improve.”

For now, most debt funds are just at the starting line. “Right now, we are operating at the task level, using AI to automate particular tasks, and we are moving towards the workflow and process level where those tasks can be connected,” says Kenny.

“Next comes the concept of AI systems, where AI tools are talking to each other to lead to the next stage of evolution. It’s about the new types of offering we might be able to develop and the new services we might be able to create, driven by client demand.”

He says that people talk a lot about the risks of AI but that he considers doing nothing and getting left behind as the biggest risk for most private credit managers. It will get harder and harder to catch up and early movers are the ones that are going to take the competitive advantages, he adds.

James says: “AI becomes really interesting when you start integrating it into the parts of the process where people do have bias, particularly where the is a heavy reliance on sponsors and some risk of over-reliance on other people’s decision-making.”

Looking forward, there is no doubt that technology will be taking on far more elements of the private credit value chain than many have yet to even imagine.

‍