AI today is ubiquitous. “The idea that artificial intelligence is the stuff of science fiction is a fallacy. The technology is now a commodity,” says Rich Klee, director of Prism at digital due diligence specialist Palladium. “We use it every time we open Google Maps, rely on predictive text or when our fridge intervenes to tell us we probably shouldn’t be snacking at 11pm.”

Public market algorithmic trading is also dominated by big data and AI. But private equity remains wary of embracing this technology and instead often continues to rely on acquired wisdom, supplemented by a touch of gut instinct.

“Given the significant human capital across the private equity industry, one would expect an abundance of opportunity to have software complete a task usually performed by a human through the use of AI,” says Andrew Tarver, founding partner and head of Motive Create, the value-creation team at Motive Partners.

74%

Private markets firms that have yet to review AI adoption

50%

Private markets firms that rate the effectiveness of advanced tech, such as AI, as medium or high for portfolio monitoring

52%

Private markets firms that rate the effectiveness of advanced tech, such as AI, as medium or high for deal due diligence

Source: Private Funds CFO Insights Survey 2022

“But that is not the case. Private equity firms have not typically been early adopters of technology, often relying upon smart people using basic tooling, such as [Microsoft] Excel and PowerPoint. Some firms have made the leap to Box or other document sharing systems, but that really is as far as it has got.”

The most frequently cited inhibitor to greater AI adoption is a lack of standardized data at scale.

“Without a vast array of underlying data in a digital format, AI does not have anything to work from,” explains Tarver. “Many would expect the private equity industry to have well-defined processes, making it easier to apply AI, yet each deal is different. Every set of quantifiable financials requires additional qualification and robust questioning. Interesting investment opportunities require a heavy dose of judgment and that is where AI is less effective.”

“One of the building blocks for the use of AI in private equity is good data collection and the standardization of metrics,” says Palladium data product manager Ryan Chapman. “With this in place you gain richer insights from the data using less technical methods before pushing into machine learning and AI.

“Private equity firms have been slow movers in this area, partly as they are reliant on data supplied by target organizations and from their portfolio companies. Data metrics are rarely standardized portfolio-wide and there is little consistency of measuring beyond pure financials, with each company treated separately. There is still a lot of foundation work that needs to be done.”

Meanwhile, Gilbert Kamieniecky, head of private equity technology at Investcorp Technology Partners, calls for a collective effort by managers to provide the necessary data to third parties that would help AI software, and then, in turn, boost efficiency. “An anonymized database across portfolios with relevant metrics could help us identify early industry trends and perform market analysis,” he says.

Arguably the greatest roadblock to AI adoption, however, remains one of culture.

Significantly more weight continues to be placed on earned acumen than on data. Until that shifts, the investment will not be made available to build the data teams and warehouses that must precede the use of AI.

AI and origination

There are exceptions of course. A small but growing number of tech-savvy firms are tentatively exploring the use of AI in private equity, and the most common use case being employed is deal origination and curation. “That is where firms are making the biggest investments in proprietary data technology, which makes sense given competition for deals,” says Klee.

EQT’s Motherbrain is among the best-known examples. “Motherbrain can track hundreds of millions of footprints, from people and their affiliations to companies and their histories, and the connections between them,” says Alexandra Lutz, head of Motherbrain at EQT. “We can ingest massive amounts of information about deals, people and companies and build useful algorithms on top of it.”

EQT then assesses companies based on the algorithms in its custom-made workflow platform, where it records its ground truths and key assumptions. “This is the hardest thing to get launched and is our secret ingredient and competitive advantage,” says Lutz. “We have created a platform where deal teams can log their interactions and decisions, and we can capture that data continuously. This connected loop means we can build predictive capabilities that get better and better over time.”

Indeed, Lutz says it is not about building a perfect model, where you can press a button and “auto- magically” find the ideal investment. Instead, Motherbrain helps teams identify opportunities earlier, build conviction faster and generate an edge to position EQT as a preferred buyer.

Having identified a target – potentially using AI tools – assets then enter the due diligence phase. Here, the use of AI remains nascent. It makes sense to employ AI to find a signal in the noise in the growing number of areas where data can be leveraged. Indeed, data availability has increased dramatically over the past decade and particularly in the past two years, as companies have pivoted their business models online in order to weather the pandemic.

Palladium specialises in due diligence of digital operating models, and its new software-as-a-service product, Prism, uses AI to analyze data extracted from a multitude of third-party data sources to determine how well a business is performing in areas, from marketing to technology to employee engagement and team growth and retention.

“AI is either going to be endorsed by the incumbents… or there will be a disruptive play by new entrants”

Oliver Gottschalg
HEC

“In this way, we are able to give a rapid assessment of a company’s performance and identify potential red flags, even while privileged inside data is off-limits to a potential buyer in the early stages,” says Klee.

A good example of how Prism intends to achieve this is by using natural language AI to analyze a target’s website and make comparisons between time periods. “We can see what skills they are looking to recruit through the job ads, and which key staff have left and joined from the team pages,” Klee explains.

Democratizing due diligence

Another interesting use case for AI in private equity involves the “explainability” of data. Private equity firms are keen to push diligence to the edges of their organizations, so that instead of lay people coming to a centralized head of digital for an opinion on a company, that capability is democratized.

AI can be employed to lead non-technical deal teams through data-intensive analysis by making data and charts explainable. “It is all very well presenting data, but if the narrative behind that data can’t be comprehended in the right way, it is worse than useless, it is downright dangerous,” says Klee.

Prism, for example, is exploring technologies such as language comprehension and generation tool GPT-3 to write automated diligence reports that sit alongside the data. “For example, for a diverse group of stakeholders an auto-generated commentary in plain language could communicate that a potential growth limiter in social marketing is a red flag and should be raised at a first investment committee meeting. If required, an analyst can then be employed to do top-up diligence in that area,” Klee explains.

Once an asset is in a portfolio, meanwhile, the emphasis turns to monitoring and value creation. Here, private equity is ahead of the game. PwC recently found that 31 percent of private equity-backed companies employ AI solutions compared with 23 percent of their non-private-equity-backed peers.

“The application of AI can be used to potentially improve efficiency, drive higher profitability, identify new sources of revenue and reduce risk, which are all common use cases across multiple industries,” says Tarver.

Kamieniecky points to Investcorp’s cloud solutions and managed services portfolio company Calligo, which stores and handles a large quantity of data for its clients. “The deal team, together with Calligo’s founder, saw an opportunity to start using AI and so employed data scientists to help companies set up learning models, clean data and provide actionable insights for their organizations.”

In terms of portfolio monitoring, Investcorp uses software to automatically monitor and benchmark across its portfolio companies, from transaction data to operational results and forecasts. “Our tools allow us to highlight if there is any deviation from our planning and help to take prompt action,” Kamieniecky explains.

Automating exits

AI can also prove valuable at exit, particularly as it is a defined process that is largely in the control of the financial sponsor. “The data required to exit can be stored in a digital format, enabling greater use of AI techniques to support the exit process,” says Tarver. “Precedent transaction tracking and target-acquirer profiling in a given space is possible. I would see this becoming an earlier AI win for the industry than the initial due diligence process.”

“We can ingest massive amounts of information about deals, people and companies and build useful algorithms on top of it”

Alexandra Lutz
EQT

“We have 40 years’ experience in private equity and as such we hold a vast data lake with hundreds of investments and datapoints,” adds Kamieniecky. “All of this information has historically been drawn from our experienced team members at IC level, but we are making efforts to generate a consistent dataset to perform structured analysis. We also have a long track record of IPOs – and again, there is vast data there that we can rely on to benchmark our companies effectively and understand how our investments could be best valued.”

Meanwhile, Oliver Gottschalg, professor of strategy at HEC Paris and head of research at MJ Hudson’s Fund Performance Analytics group, is particularly interested in the use of AI from the perspective of the limited partner.

“How can AI be used to identify attractive primary fund propositions? Or to price a secondaries stake? Or to determine what makes a good co-investment deal? These are all the things that a big fund of funds player would focus on, but in this case, determined not by human judgment, first and foremost, but by using AI,” he says.

“The initial results we have observed in this area have been phenomenal. A massive performance improvement can be obtained, not by switching off the human brain entirely, but by leveraging AI tools to carry out the first layer of analytics that can then be supplemented by human judgment on governance, strategy and people.”

There is undoubtedly potential for AI to be additive throughout the investment life cycle. How profound that transformation will ultimately prove remains to be seen. But what is clear is that private equity firms cannot ignore the dominance of data and the tools for leveraging its power for much longer.

Technology will always follow the money and vast sums have been flowing into private equity over recent years. If that continues, it seems inevitable that big tech companies, or hedge funds armed with teams of technology-infused quants, will make a move on the asset class.

“Some form of algorithmic private equity firm that places a bias on data over subjective business acumen, fireside chats and handshakes is bound to emerge,” says Klee.

“AI is either going to be endorsed by the incumbents – traditional firms will reinvent themselves – or there will be a disruptive play by new entrants,” agrees Gottschalg, who recalls a recent conversation with a board member of a large European fund of funds who was concerned about whether the firm would have any basis for charging management fees within a decade given the risk that one of the tech giants could blow them out of the water.

Indeed, it is theoretically possible that a completely hands-off investment cycle akin to algorithmic trading could emerge – as anathema as that idea would be to the majority of private equity practitioners today.

“A private equity bot would be programmed with investment criteria, or to put it another way, a firm’s DNA would be expressed in code,” says Klee. “That bot would then systematically select deals and build the investment committee case. Clearly, there would still be a human rapport angle, with deal teams responsible for building relationships with the management team. Everything else, however, would be defined in code and decisions strictly weighted based on the available data.”

A private equity industry dominated by bots may seem fanciful. The asset class isn’t going to forgo the cult of the business guru any time soon. But there is no doubt that the world is changing. Data is the currency of business and firms that fail to establish the culture, processes and tools to leverage its power may be left behind.