Cloverlay is not your typical mid-market firm. It is young – founded in 2015 – and bills itself as an investor in “adjacent private markets” through co-invests, platforms, joint ventures, fund restructurings and secondaries. Like many private capital firms, however, Cloverlay is a “slim shop,” says principal and CFO Omar Hassan. The firm has 12 full-time staff managing approximately $360 million. It will soon raise Fund II with a $400 million cap, according to market sources.

Hassan was previously a controller at Apollo Global Management and held similar roles at other firms in the credit and PE spheres. “I don’t have a tech background, but I have always dabbled in it,” he tells Private Funds CFO. “Robotics is taking operational efficiency to the next level.”

Processing invoices used to be a thankless job that required a lot of human hours and “you really needed to get it right,” says Hassan. Thanks to robotic process automation (RPA), for the last two years this has not been the case. Supplier invoices are now sent to an email inbox at Cloverlay; the invoices are scanned, information extracted and costs allocated to the appropriate funds or cost centers for sign-off.

This was never a massive volume task for Cloverlay; the firm processes anywhere between five and 15 invoices per week. But the human intervention typically required to allocate the expenses with consistency meant headaches and human error. The robot, which has been taught how invoices from various suppliers should be coded, now takes the strain. Humans are not removed entirely from the process, but by the time a human is involved around 85 percent of the work is already done.

It took Cloverley between four and six weeks to get it up and running and complete the necessary testing. The firm worked with a consultant; “in the grand scheme of things it was pretty painless,” says Hassan, who has discussed this with other CFOs at similarly sized firms, but has yet to come across a peer doing the same thing. A lack of volume is a commonly cited reason, he says.

If you’ve not come across RPA before, it is – put simply – a way of getting a piece of software to undertake simple rules-based routine activities. Think of a program that can open emails, access databases (it has its own log-in, like a human), gather data, fill out forms and perform calculations. It is best suited to repetitive, rules-based tasks that involve structured data. Among the leading providers of RPA software are Blue Prism, Ui Path and Automation Anywhere.

Giant financial services firms have already caught on. In the space of 20 months, Royal Bank of Canada’s wealth management division used Blue Prism software to save 153,000 hours of direct manual work, which in turn equated to circa 170,000 hours “given back” to the bank (the extra 18,000 hours comes from not having to correct errors). One process that formerly took a human around six hours is now completed in around 10 minutes by Blue Prism’s bots.

Contract killers

AI’s ability to read, translate and determine sentiment from documents or human speech can revolutionize contract analysis. EisnerAmper uses IBM’s Watson for its AI solutions; a contract that could take a human staffer two to three hours to review may only take Watson a few seconds.

“It can ingest the contract and then produce a summary of important issues, such as revenue recognition trigger points, along with any other issues that auditors and internal accountants identify,” says EisnerAmper’s Jay Weinstein, managing partner of markets and industries. He also suggests that this type of contract analysis could work wonders with company pension plan documentation; those 400-page doorstops. AI can now be trained to look for key matters in those documents.

“For the finance function specifically, machine learning through software functionality will automate bookkeeping transactions as they happen across bank accounts,” says Jason Bingham, managing director of product development for the fund administrator Sanne. This means recognizing transactions, even new transactions, posting them, and with machine learning, classifying that data and pulling it through to general ledgers.

Most private equity firms are not of the same scale of RBC. This should not preclude them employing RPA within their operations, as the Cloverlay example illustrates.

At the private equity coalface, however, “many smaller PE houses operate on spreadsheets rather than purpose built systems,” says Ben Booth, chief information officer for fund administrator Ocorian.

We asked 15 CFOs of mid-market private equity firms how many used an automated waterfall calculation tool, and – in support of Booth’s assessment of the situation – only three said they did, with respondents saying their waterfall was either too simple or too complicated to automate.

But the industry is changing. What could be called enterprise resource planning solutions for private equity – systems provided by the likes of eFront, iLevel and Investran among others – are starting to proliferate.

“The ability to turn this data into usable information and the speed in which you can do that is transformational,” says Richard Butler, the COO of lower mid-market firm ESO Capital. “We’ve recently implemented a powerful platform called Ipreo, enabling us to better compare our portfolio company KPIs, even for businesses across different industries.”

For the last three years, Harbourvest’s CFO Karin Lagerlund has been working on upgrading her team’s processes. “We looked at where we had cumbersome manual processes, or a lot of Excel spreadsheets, and began prioritizing how to digitize and use technology to streamline those activities. We digitized our journal entries and created a system that pulls the back-up data automatically and stores it for us in a cloud solution,” says Lagerlund.Butler explains that just three years ago, he would need a team of people to manage this information, and now he can do so at the touch of a button.

Automation will pay a huge part in the next generation of improvements. EisnerAmper’s Jay Weinstein, managing partner of markets and industries, explains how his firm is channelling resource into producing ‘bots’ that are capable of automating major processes within PE firm operations.

As an example, a fund may have month-end reporting for thousands of investors that they are currently recording on Excel spreadsheets. These spreadsheets include numerous calculations, which they spend hours producing. A bot can automate the spreadsheet and eliminate the manual inputs and reconciliations to validate those investor calculations.

“Those tedious reconciliations can take two or three days to complete,” says Weinstein. “And with the bot, the staff is freed up to handle more interesting and productive work, while also enhancing accuracy by limiting input errors.”

These technologies allow firms to migrate to an exception-based review and
reporting process, so that staff are not looking at every number or calculation, only those that represent aberrations, say when a data point communicates a loss, when the market for a particular portfolio company is booming.


RPA may not be for every firm. Partners Group – a private markets firm with $83 billion in assets under management across multiple asset classes – has a 90-strong technology team. The firm recently worked with a consultant to analyze the firm’s operations to see where RPA could drive efficiency. The result came up negative. “Because we are quite tightly integrated from the front office to our database, we did not find a single place where we could get upside from RPA,” Raymond Schnidrig, Partners Group’ chief technology officer, tells Private Funds CFO.

“A typical application is when reconciliation is needed between two systems that are not talking to each other. We don’t have much of that.”

Partners Group does deploy software robots to conduct regression testing when implementing new software. “We have quite a big bank of internal developers either building or buying in software,” says Schnidrig, “We have to test this and use robots to do it, which is quite standard I believe.”

What would it take to make RPA more relevant to Partners Group? “The heavily rules-based RPA methods would have to become more intelligent, with more variable input and output handling; this could result in coverage of more use cases,” he says. “Then it might become more applicable to lower volume work.”

Which leads us on to artificial intelligence. There is much excitement around how AI will change private equity firm operations. A survey conducted in late 2018 by fund administrator Intertrust found that 91 percent of private equity professionals believe AI will disrupt their sector within the next five years (the firm surveyed 80 PE executives).

So how does artificial intelligence differ from RPA? Definitions of these types of tech can be slippery (and debated fiercely by technologists). Consulting firm CBF Bots puts it like this: “On the most fundamental level, RPA is associated with ‘doing’ whereas AI and ML is concerned with ‘thinking’ and ‘learning’ respectively. Or brawn versus brains, if you like.”

Taking the invoice processing example: RPA knows what steps to take, because it has been told, but it may require AI to read the invoice and consider where to find each bit of the data it needs.

By this definition, AI is already hard at work in many private equity firms. Anyone using an expense system that scans pictures of receipts taken with your phone and transforms these into a claim – such as SAP Concur – is already working with it.

AI is also more often characterized by its ability to take oceans of data and discover patterns that humans cannot. And this is where private equity firms are starting to use their imagination.

Returning to Partners Group: the firm worked with a niche provider to develop an AI approach to identify negative news about its portfolio companies. This resulted in the creation of a bespoke tool used by its ESG team as part of its due diligence and portfolio monitoring activities. The firm has several thousand portfolio companies around the world, so to use conventional search engine notifications would have produced too much “noise” for humans to sift through. The AI program does the sifting.


The Securities and Exchange Commission has understood the power of data analytics in its work for more than a decade. It has been conducting deep data analysis on the public company front, particularly using XBRL, Extensible Business Reporting Language, for many years. XBRL, which is based on XML language, is a way to exchange information and data between different systems. It already is starting to use machine learning for reviews of public companies and for detecting potential market misconducts.

Much of this data-driven approach work has focused on publicly traded companies, and it remains to be seen whether the SEC will duplicate that to private equity firms and whether it can use some of the same analytic tools to perform exams of general partners.

With the exams being locally run and the SEC putting such an emphasis on visiting firms and staying weeks to analyze documents, it’s unlikely the exam process will become much more automated in the next few years.

“SEC inspections are likely to increasingly leverage access to data and systems, yet personal interactions will probably continue,” says David Larsen, a managing director at Duff & Phelps.

Deal sourcing sorcery

Most private equity firms do not have many thousands of portfolio companies to monitor. Indeed, most do not even have ESG teams. What all private equity firms have, though, is a need to source deals.

A million analysts to look at a million companies? How about one investor to look at a million companies? This is the vision presented in a slide deck from Paris-based growth equity investor Jolt Capital.

Jolt is a relatively new firm, having been established in 2012 with backing from Temasek subsidiary Vertex Ventures among other investors. It has a team of 11 and right now it is raising its fourth fund for which it is seeking around €200 million, according to market sources. It is not a private equity giant, but the expectation of its partners is that one day it will be.

Jolt’s ability to scale should be greatly enhanced by its investment in a proprietary AI-driven software platform, Jolt Ninja.

Ninja monitors a vast array of opensource and third-party information services to find suitable investment targets. Since starting development in 2016 and going “live” in 2018 it has led to two bolt-on acquisitions for portfolio companies and one new investment for the firm. Another Ninja-sourced deal is due to be announced imminently, managing partner Schmitt tells Private Funds CFO. This is impressive for a firm that is not a volume player.

“We invest growth capital in late stage tech companies with revenues of between €10 million and €100 million,” Schmitt says. “These are often too big for TechCrunch to cover and too small for the Financial Times. They are not sexy.”

As a result of this lack of media sex appeal, Jolt’s target investment universe is opaque and not easy to map; at least not for the human brain. Ninja, however, ingests data from open web sources, like company websites, as well as closed ones like subscription databases. It takes in datapoints from LinkedIn, news sources, patent filings, events and other data sources numbering in the hundreds of thousands. It uses natural language processing software to extract relevant information and build graphs of people, companies, revenue numbers and distribution agreements with resellers.

It has ingested and processed information on over 300,000 companies. Now half of all Jolt Capital’s outbound deal sourcing is led by Ninja’s recommendations.

Ninja even tailors its recommendation depending on the preferences shown by each member of the investment team.What’s it like working with this type of robot? Much like working with a rookie human who responds well to training. “Every week I get an email saying that here are the companies that have been matched by Ninja,” says Schmitt. “At the start, probably only one in 20 leads would be interesting, but the more I tell it what I like, the better it gets.”

Such a program is not cheap to build. Schmitt does not give a figure for the investment, but the project has been the work of three full time engineers reporting into chief technology officer Philippe Laval and four analysts since work began on it in 2016.

Exciting as it is, has it been worth it? If Jolt were to remain a small player – raising less than €100 million per fund – then the answer would be no, says Schmitt, but this is about scalability: “We formed Jolt in 2012 and are growing fast. We asked ourselves what firm we wanted to be when we reached €1 billion in AUM. We are now more scalable.” Ten percent of Jolt’s revenue will continually be invested in Ninja.

Side note: the existence of Ninja has, says Schmitt, has had a deep impact on the partners investing behavior; they feel they can negotiate on a different footing. “Most people are worried about losing an opportunity. We are less worried about losing a deal because our dealflow is so robust.”


Heramb Ramachandran is CFO of XPV Water Partners, a specialist private equity firm that invests in water-related companies. It has $400 million in capital under management and an operations team of seven.

What does robotic process automation mean to you?

I view it as breaking down a process into its respective components and determining which components can be performed by humans or machines.  Through process automation, the ultimate goal must be to simplify the original process while reducing risk.

As a firm believer in lean back offices, the benefits of process automation in the finance world are evident. For example, a CFO’s period-end closing of the books can be potentially automated. Many of the components of this process are routine, high volume and can easily be performed by a machine. 

How about AI? Is that on your radar?

It is certainly on my radar. I am continuously evaluating strategies to drive value for an organization while managing risk. With the evolution of AI, there will be an even greater emphasis on data-driven analysis that will require the back office to adapt. The days of plodding through spreadsheets to generate a calculation will soon be replaced by an AI algorithm.

Have you implemented anything recently that has reaped efficiency gains?

Sometimes it’s the simple things that generate the most value.

We introduced Expensify, which is a cost-effective, cloud-based expense management system to our organization. We used to prepare expense reports in excel and manually post the individual transactions to our general ledger. This was time-consuming, prone to error and added zero value to the organization. There would often be a one-month lag before an employee was reimbursed their expenses.

The introduction of this simple software has transformed a process that previously took the finance team two days to complete down to 30 minutes through effective process automation.  Moreover, employees were reimbursed within one week of submitting their expenses, so the benefits of this software impacted the entire organization.

How long will it be until the finance function is staffed by robots?

Financial operations lag behind other functions with respect to widespread adoption of innovative technologies. The majority of back offices at small and mid-size companies use outdated accounting software with fragmented technology stacks. Reconciliations performed outside the system still tend to bridge the data across multiple platforms. There are years of process inefficiencies embedded in the operations which can’t be fixed overnight. The full-scale adoption of AI and process automation are still years away to practically benefit smaller organizations.

Robots finding deals

The ability to find and assess investment targets is clearly where many firms see the future. Returning to the Intertrust survey, a third of respondents identified “screening investment opportunities” as being the area in which digital innovation is currently having “the greatest impact,” but nearly two-thirds (59 percent) said it that this would be case in five years’ time. Likewise for “quicker due diligence” and “improved investment decisions.”

While a young firm like Jolt Capital is looking outward at what data it can absorb, firms with decades of their own data are looking inward. One is Riverside, a global firm focusing on businesses with enterprise values of less than $400 million. In private equity terms – certainly in the lower mid-market – Riverside is a volume player, with more than 90 companies in its portfolio. The firm has made more than 600 investments since it was founded in 1988, so within its vaults is an ocean of data to analyse.Before this dream becomes a reality, Riverside must overcome an obstacle that will be familiar to many of its peers: organizing and structuring data that is currently “scattered around a dozen or so” different internal systems. “Don’t underestimate the data management piece,” says Feldman, “getting the core plumbing in place in order to leverage the data – ours and third-party data.”Chief information officer Eric Feldman tells PFCFO that the firm’s leadership dreams of a presence on the investment committee akin to IBM’s Watson, the computer system made famous by winning the TV gameshow Jeopardy! in 2011. Riverside’s version would be able to look at a small manufacturing company in Omaha and immediately analyze hundreds of thousands of comparable businesses around the world in a way that none of the humans around the table could.

Implementing the AI technology is the easy part, experts say. What’s difficult and can turn firms off the whole idea is the creation of one data source. Before firms can take advantage of AI technology, automation processing and robotics, they have to go back into their data and create a data warehouse.

“Surprisingly, the single ‘source of truth’ approach to deal-management enabled by effective data warehousing was not adopted by a large margin of funds until very recently,” says Michael Asher, chief information officer of RFA, a service provider with a focus on technology. “By doing so, GPs can build algorithms that look at portfolio data from all phases of the cycle holistically and say, ‘Well, this is a great deal’ or ‘Our research indicates this is something that we should be looking at.’”

In other words, AI requires an enormous amount of data to learn from, says Jason Bingham, managing director of product development for the fund administrator Sanne. “To be effective, machine learning needs millions, if not billions, of data points,” says Bingham. “And many firms are still in the process of structuring all their data into a ‘single source of truth.’ Much of the ‘big data’ captured in the past five years does not as yet have a long enough time series to be properly validated and may be more commonly used by firms five years from now.”

Luddites will be proved wrong

So we are all agreed: robots are great. But how afraid should members of the knowledge workforce be as automation eats into their workload? A recurring theme among CFOs and service providers is that this next wave of technology is about maximizing what current staff can do. The asset class tends to staff leanly as a rule and every firm aims to do more with less.

“We are looking for how many hours automation can save, not just now, but during future growth as well,” says HarbourVest’s Lagerlund. “So, we fully expect to be able to increase efficiency, without necessarily increasing our accounting staff.”

Cloverlay provides a case in point. Processing invoices was never a full-time job – two or three hours a week typically. Instead, the hours “given back” allow operations professionals room for creativity. “It allows us to step back and say, ‘What do we want to spend our time on?’” says Hassan.

Some firms are hiring a special class of IT project manager – with skills in both IT and project management – to make the most of today’s solutions. They collaborate with business users and the IT team to decide if a project could be handled by a consultant or in-house, and whether a solution is available off the shelf or would need to be built from scratch. And then they see the project through to completion.

For firms with less tech in their DNA, the revolution may lead to their first CTO, or a different expert altogether. “Data management is only going get more critical and complex,” says Lance Taylor, CFO of Palo Alto-based mid-market firm HGGC. “So, there may be the need for a kind of data scientist, or data expert, who can think about how best to organize and manipulate data for value.”

For many it is difficult to imagine a world in which robots take on human tasks and the result is not human job losses. But as ever, the picture is more complicated than that.

For small, young, lean firms with ambition, automation provides tools to scale quickly. In the case of giant firms with legacy systems, we are more likely to see repetitive finance jobs replaced with technology positions.

We have grown used to the idea that the role of the private fund CFO transcends finance to encompass wider operations and projects. In the automated world this will be more so than ever. The future-proof CFO will be adept at managing IT projects… or hire someone else who is.