Private funds, while often enablers of operational efficiencies, are now starting to look at their own operations to find areas ripe for disruption.
The last decade has been described by some as the “golden age” of private equity – conditions didn’t quite demand an efficiency mindset at the firm level.
But times have changed – the era of high valuations, lower leverage costs and minimal interest isn’t likely to be coming back any time soon.
While it’s not the only emerging tech we’re going to discuss today, AI is certainly top of mind for most of us. We’re racing to understand the impact that recent advances in AI could have on portfolio companies, but most firms have made only tentative progress in using AI tools to improve their own operations.
AI’s potential, while vast, is still at early stages – effectively applying the tech while things are changing at pace makes implementation a challenge, and AI takes time to deliver significant, tangible value.
Reliability and accuracy are both significant sources of frustration when it comes to AI, particularly at the portfolio level. A thorny challenge is bias – AI leverages human generated data, so it perpetuates human bias when analyzing this information and has been widely known to fabricate answers: evidence of which is so prevalent that it can’t just be excused as an outlier.
Risk governance is also a consideration for firms wanting to leverage AI: allowing the software to access proprietary data could erode the firewall they use to keep that data away from competitors.
However, as we’ve started to see from very early success stories among the biggest players across the industry — like Blackstone and Carlyle — time and training will likely smooth over some of these issues. But it’s like someone has set off the starting gun on a massive global race: however it’s a marathon, not a sprint – over time we’ll see clear leaders emerging as innovation journeys play out.
Still, conversations we’ve had with a range of private markets firms reveal many are in the pilot phase of implementation, experimenting with quick wins and efficiencies that can boost productivity. The higher pursuits of gleaning more intelligent insight through AI remain, largely, in the distance.
Most of us agree the use cases of AI are numerous. Among the more commonly anticipated impacts is that it will level the competitive playing field – generative AI’s penchant for analytics at speed is expected to challenge the information asymmetry that forms the private equity competitive edge.
But if we’ve learnt anything from the very public, often comical and sometimes disturbing experimentations with AI this year, it’s that throwing data and AI into a pot and hoping for the best doesn’t always work out. Models need to learn, not only to avoid bias and false positives, but also to understand the user’s specific needs.
Watch the full video below and read on for a summary recap.
Danny RIcciardi:
Hey everyone, my name is Danny Ricciardi and I'm the CFO at Centana Growth Partners. I've been with Centana for a little over seven years now. Prior to Centana, I spent some time in public accounting at a family office. Centana is a growth equity firm that focuses on the financial services, fintech and enterprise technology sectors. It was founded in 2015, and currently we are up to 24 portfolio companies and a little over $1 billion in AUM. As a small firm, we're only 20 people. My role at Centana covers many areas. My main focus areas are financial tax reporting for our funds and management company, IT legal and compliance and vendor management.
Sar Ruddenklau:
You've got a unique position where you see your portfolio companies that are using tech and AI, and you're driving it within your firm. What do you think has been the most impactful digital tool for your fund operations in the past few years?
Danny RIcciardi:
A couple of years ago, we implemented a portfolio management software, which has been a very successful tool that helps centralize the KPIs for all of our portfolio companies. Everyone at the organization has the ability to log on to a centralized database and see summary reporting at the portfolio company and fund level, for inception today in all of our portfolio companies. As well as it's also streamlined our internal quarterly reporting process on the portfolio company. The system also helps generate the reports that we use for discussion during those meetings each quarter.
Sar Ruddenklau:
Joe, over to you. Can you introduce yourself and give us a brief description of your role at GP Fund Solutions?
Joe Capobianco:
Thanks, Sar. I'm Joe Capobianco. I'm the director of our data group at GP Fund Solutions. For anybody who's not aware of GP Fund Solutions, we're one of the boutique fund administrators, primarily working with closed-end funds across a range of various strategies, including buyout, growth, VC, debt, real estate. We've even worked with some family offices standing up an institutional reporting platform for their private market portfolio. We're headquartered in Latham, New York, which is about a two and a half hour drive north of New York City. We have various satellite offices around the country, Minneapolis, Boston, Fort Lauderdale.
Over the past couple of years, we've been standing up our international arm, which is in Brighton, UK and Limerick, Ireland. My team specifically started in 2019, which is when I joined the firm. It was really to help align GPFS around data, and align the firm around a consistent, standardized reporting deliverable. Also, expand our service offering to clients.
Our primary objective on the data side is to create a cross-functional data structure that allows for accessible and usable data across your firm's teams and various reporting requirements you may have. The primary delivery method is through customized dashboards and reports that my team builds and manages.
Sar Ruddenklau:
Again, you've got a fairly unique perspective seeing private equity firms using tech at scale. What, in your opinion, has been the most impactful digital tool in fund operations that you've seen over the past few years?
Joe Capobianco:
I think the biggest thing that we've seen, and we'll get into this a bit later, is introducing AI and technology as they’re looking to move out of Excel, which has really highlighted the gaps and the reporting deficiencies within the firm.
Everybody's trying to get their hands around that and figure out what the best way is to centralize this data, to get a consistent data source that everybody's drawing from, so that's really been my focus over the past couple of years.
Cassie Gharnit:
What were some of the major challenges that private equity firms were facing in their operations before the advent of these more advanced technologies and AI? How about we start with you, Danny?
Danny RIcciardi:
I think it's a broad perspective, but for my specific use cases, I feel like data was very siloed on people's networks living on drives and various Excel files across the network. There was always this need to have it centralized so people in real time could have access to it, as opposed to emailing one-off questions.
That led to inefficient workflows where an LP sends a request to the IR team. The IR team goes to the finance team, they send the number. Instead of having this data live in a centralized location, all this stuff had to be parsed together by different parties at the organization to complete things that everyone at the firm should have access to.
Joe Capobianco:
I'll definitely echo that sentiment. I think what we've seen is a proliferation of technology that looks to solve specific issues or specific use cases, without really addressing the root cause of how we can get the firm aligned from a data perspective. How we're approaching data, how we're producing data, consuming data, processing data, in order to get all these various teams that Danny's referring to aligned across the firm. The idea is you have one source that can be cut or split multiple ways for all the different use cases.
From the standpoint of having a debt strategy, an equity strategy and maybe something else: in order to get an investor cut of data across all those various strategies, you're taking data from various Excel files, combining it together. You're assuming that everybody looked at terminology or is thinking about things the same way. In order to normalize that, that's a giant effort that the firm has to go through, when it should just be the click of a button from one of the IR individuals. That's what we're coming out of right now, and the question is how are we going to get the firm aligned? That's where we're at today.
Sar Ruddenklau:
I want to ask you about the unexpected consequences and challenges that arose from integrating tech into operations. Maybe it's because they’re just point solutions, which was the primary model of these early tech solutions. But what else is going on there? What were the biggest challenges?
Danny RIcciardi:
From my perspective, the biggest challenge is bandwidth: the allocation of time and resources. We have a small team. There are a lot of great technologies out there that I'm sure would help make things efficient.
But it's important first to get full firm buy-in on that technology or adoption to make sure interests are aligned. Because it is quite an effort for smaller teams to implement these technologies, and you want to make sure that you're using resources cautiously.
The other side of it is the actual cost of the technology. LPs are very cost conscious of their expense ratios at the fund. You need to make sure that any technology that you're adding is actually providing value to them at the end of the day, and it's not just another expense that they're incurring.
Joe Capobianco:
We're taking a very bespoke, nuanced dataset that each firm is dealing with individually, and we're trying to fit it into a standardized structure. That transition is, I think, what limits a lot of firms. Then the adoption curve, of course, can this technology do everything that we're asking of it?
As much as we say we want to move off of Excel, one of the big things that makes Excel great is the flexibility. I can just take a cut of data, I can add a couple columns, add a couple of rows, and really slice it instantaneously. I think that's why Excel is that sugar high, where it's just so easy to produce what I need.
Whereas if we're thinking about integrating an application into a firm, we really need to think through how we are structuring the data? How is this data connecting with other systems? What types of sources is it feeding? What is it ingesting data from? That takes a lot more effort, but it's definitely more sustainable in the long run.
Sar Ruddenklau:
We've talked a little bit about the past. So bringing it into the present, most private equity firms have adopted tech on some level, whether it's Excel, whether it's something a little bit more advanced, whether it's a point solution or something more end to end. What do you think the key drivers are behind this adoption?
Danny RIcciardi:
As a tech-focused firm, we want to make sure that we're presenting ourselves in a very similar manner in terms of being up to date and being in tune with all the stuff that's going on. We have to adopt all those table stakes technologies to present ourselves in a tech-forward way. The other side of it is regulation. Increased investor demands lead to increased reporting, which increases the need to have some sort of system to generate those reports.
The private funds rule, which is current events, was actually reversed the other day. We’re all still evaluating the consequences of that, but that was big news in the industry. For folks who haven't seen that, that was appealed yesterday so as it stands, all those things that were going to go into effect in the fall are no longer required. But that kind of regulation is what is driving some of this adoption.
Joe Capobianco:
The transparency demands around LPs are not going to decrease regardless of the SEC, so it's still something that we need to think about in terms of future proofing. Even ILPA is rolling out a whole new set of regulations. Is the data structured in a way where I can just efficiently make an update to my dataset? Maybe I can add an attribute or repurpose something that's already existing to meet a new reporting requirement. Obviously, you can't think of everything that's going to happen. But if structurally you've created that dataset where you can easily manipulate it or easily pivot to a new regulation, that's going to be key.
We're this far along in terms of increasing transparency. Institutional investors are only going to want more information around the data and around the reporting. So even though the SEC decision was a big one yesterday, it's definitely still top of mind for institutional investors.
Cassie Gharnit:
In the spirit of keeping up with the transparency with all the different changes going on, we're going to chat about the AI elephant in the room. How do you see private equity firms leveraging AI and how is it shaping operations?
Joe Capobianco:
From the industry conversations we've been having, nobody wants to be a first mover but nobody wants to be left behind either. When we think about AI, LLMs really made AI the buzzword over the past couple of years. But from a machine learning, deal sourcing standpoint, that's been in play for a while now.
Once ChatGPT rolled out, it caught everybody's attention and all of a sudden, AI was this huge deal to everybody. I think we should make that distinction in terms of what's the machine learning side? What's the LLM side?
On the back-office side, there's definitely a huge opportunity, but the data isn't good enough to capitalize on it. I don't think the LLMs are there yet either in terms of producing a quality output. I can't just query a database, have the LLMs search the entire dataset, figure out what I'm looking for and return a 100% accurate result. I think that's probably the direction that the industry's going, but we're just not there yet.
Danny RIcciardi:
Joe, your initial point really resonated that we don't want to be the early adopter but we also don't want to fall behind. That's exactly how we're viewing it here at Centana. We're definitely, carefully evaluating the best way to integrate it into our operations, aware of the potential benefits, streamlining processes, et cetera.
But we need to be careful of how we adopt it. We need to be aware of the concerns about the confidentiality of the data and make sure that it's not just going out into this abyss, and not knowing where your PII is being stored. That's obviously a critical thing that we're evaluating. Our investors ask us how we are using AI for that, and that's really at the firm level.
We have a couple people test piloting Copilot to see if we can integrate that, but it's still early on. We like the idea of using that Microsoft product just because it integrates nicely into our ecosystem of tools we already have. We also know that they'll probably have some robust monitoring tools that our IT firm can help us make sure that we're not doing anything that we shouldn't be doing.
In addition to that, we have partnered with a lot of different service providers that serve different functional areas that are leveraging AI at their organization. For example, we use some expert networks, AlphaSights, and they help us synthesize conversations that our firm has with expert networks.
We are definitely using it, our firms have already vetted it out and internally, we're carefully vetting it to find the best way to adopt.
Sar Ruddenklau:
Yeah. To your point, nobody wants to be the first mover but nobody wants to be left behind. Passthrough produced an AI and tech report earlier this year. The conversations I had with all of our customers, who I wanted to feature in the report, took the same pattern: we're not really doing a lot with AI yet, but really interested in what everyone else is doing.
I think everybody is in the starting blocks at the moment, but ready to go. Some great implementations from large, well-resourced firms that we came across were EQT's Motherbrain. EQT built a model whose goal is to help their venture team spot the hidden gems. They've had a reasonable amount of success with that as well. Blackstone is also making some serious moves in a couple of different areas of AI.
Are there any other successful implementations that you guys have seen? What is the best way to ensure success with adopting any sort of tech product?
Joe Capobianco:
I'm obviously a little bit biased, but I think the foundational piece to any type of AI is good quality data. Everybody knows the phrase garbage in, garbage out. AI just gets garbage out faster. We need to make sure that you have good quality data, and it's set up in a way that can return a good result for you. But privacy is definitely top of our mind.
You can't be involved in data and not be concerned about privacy with AI, and they are two competing items. A lot of the big players, Danny, you mentioned Copilot, Google has Gemini for Workspace, ChatGPT Team and Enterprise are all gearing towards the corporate-type environment and really trying to cater to data privacy. Especially around GDPR or any of the new state regulations that are coming into play here.
We're also tracking that, we're testing it. We've started ChatGPT Team internally here to test how it's working, using it as an assistant, and really just almost like a Copilot. But in terms of actually processing data or using client data, we haven't ventured into that yet.
Sar Ruddenklau:
Are you building guardrails around your ChatGPT use? What does that look like for you guys?
Joe Capobianco:
There has to be guardrails around it. Data governance is huge. On ChatGPT Team, it's an isolated sandbox environment. It's not being used for any type of training purposes, so we're definitely conscious of it. We're conscious of who has access to it, who can gain access to it. If somebody leaves the firm, what can they take with them? Those are all things that we're currently vetting out as we're experimenting.
Danny RIcciardi:
To that point, a big part of it for us, is ongoing training. We have a robust training, compliance training program, which obviously we've found ways to incorporate training aspects of AI. Awareness is a big part of it, and making sure people are trained how to properly navigate it correctly in a secure way is an important part of making sure you're using it efficiently.
Cassie Gharnit:
You both mentioned a little bit about some of the obstacles being the LLMs aren't always 100% accurate. There's some privacy issues. But I'm just wondering if there are any other common obstacles that you see private equity firms facing when adopting and integrating new technologies, whether it's AI or some other advanced technologies into their workflows today?
Danny RIcciardi:
I'm taking a step back and maybe separating it from AI just in terms of obstacles for implementing. In general, two factors are important: the resources, making sure you understand the full scope of the implementation.
I know you both work at Passthrough and I'm sure you express time for implementation, but I always add 2x, 3x, to make sure you have enough time allocated. The last thing you want to do is sign an annual contract and have something sit there because the implementation stalled and you don't have the time to dedicate to it.
Also make sure you have an owner of the project or a quarterback of the project, someone who's held accountable for moving it forward. Otherwise, it just may sit there and never get pushed forward. The other part of it is making sure you have adoption from the people who you work with, depending on who the technology interacts with inside the organization. It's a political game I play when I try to adopt a new technology, but you really need to talk to everyone and understand how they use it and what concerns they may have for using it. Because it's a big effort not only to implement, but to maintain it going forward.
The last thing you want to do is have that expense, spend all that time and not have it used. Once you understand all those factors and you get alignment, you’ll be able to understand if it is something worth pursuing as an organization.
Joe Capobianco:
We mentioned alignment a couple times and it's definitely easier said than done. One of the things that we've seen be successful and one of the things that we've done internally, is start a data strategy within the firm. The data strategy really helps align the organization in terms of everybody's pulling in the same direction.
They understand how their specific activities within the firm are impacting the data downstream for the consumers of the data. Something that the finance team does could impact IR, it could impact compliance regulatory reporting. Understanding that connectivity within the organization is key, and so that's really what we think about as alignment.
We established a data council within GPFS. Then we've also seen other clients think about or have implemented a data council themselves. Eventually, I think once you get big enough, you probably need to think about building out your own internal data team.
But that's really the genesis of where the GPFS data team came from. We were that outsourced data team to our clients, who maybe it didn't make sense to hire a full-time resource there. But, Danny, I think you nailed it. It's adoption within the firm and the alignment, but the data strategy puts the framework around that alignment.
Cassie Gharnit:
Looking ahead to the future now, what ways do you see technology changing the competitive landscape within the private equity industry today?
Danny RIcciardi:
There are a lot of new technologies like e-SubDocs, just to name one, that are coming out that are presenting a more efficient way to do processes that were traditionally done manually. My job is to evaluate which ones make the most sense to implement, because there's a lot of great technologies out there.
I mentioned e-SubDocs, and there's also firms out there who are doing similar things on LPAs and side letters for monitoring. There's automated diligence firms. We just went through a fundraise. There's a lot of firms that take your diligence questions that you answer, and put them into a bank and use AI to give suggested answers.
There's firms out there that automate complex workflows like waterfall management fees where there's different class structures. It's great that we have all these firms out there competing and providing great technologies. My job is to evaluate and see which ones it makes sense to adopt.
Joe Capobianco:
Going back to the Passthrough link and e-SubDocs in general, you need to think about the data life cycle throughout the fund life cycle. The first gathering point of data is during fundraising, at the SubDoc level. So being able to capture a clean dataset on fundraising will carry through the firm.
Then once you raise, let's say you're on fund one, once you go for fund two, fund three, you're making sure that that data is consistent across the firm. I've seen countless times where fund one has an investor classification a certain way, and then fund two has a different one. Then we're constantly chasing, "Well, which classification is it?" and we have to go back to the investor.
Having that formal process in an e-SubDoc application upfront will identify that difference immediately, so you know you're getting clean data on onboarding. It saves a lot of time from going back and redundant work or data clean up later on. What it allows you to do is a lot of forecasting insights around your data too. How much time did an investor spend on their SubDoc? Where did they get tripped up? You're getting insights at the very start of your fundraising, and then you're using that data to make informed decisions later on. We didn't even talk about the use cases around having a quality dataset. A lot of what firms start with is just backwards looking. What's my performance? Going back to previous reporting periods, things like that.
Then you can take that and build data science around it, for example forecasting cashflows. LPs always ask, "When can I expect to receive the first distribution?" You can actually make an informed decision around that if you have a meaningful dataset there. That's why we're so passionate about alignment around the data from the very start.
Sar Ruddenklau:
I'm going to change tack a little bit and talk about some ethical or regulatory challenges that firms possibly should anticipate. What do you see coming down the pipeline as they implement this more advanced technology?
Joe Capobianco:
One of the big topics around AI, the data centers, the processing, is how much energy these models are using. There's a whole ESG component around it. There's definitely a push within the industry to try to find a more sustainable energy source than the perceived dirty energy that was previously used. There's a push towards nuclear energy, nuclear startups powering these data centers, which is early days, but it's definitely an interesting conversation.
Danny RIcciardi:
To add to that point, Joe, our investors are definitely concerned about ESG-related issues as it relates to all matters, specifically this. Of the questions we get asked from our investors on an ongoing basis, ESG-related ones are probably the highest ones we receive.
I haven't seen questions quite yet evolve to repercussions relating to AI. But it's only a matter of time before it becomes a highlight of limited partners as firms start adopting this more. It’s definitely something to be aware of as our investors care about it a lot, as do we.
Sar Ruddenklau:
To the extent that you can share, what are they actually asking? Is it more around data governance? What's their focus at the moment?
Danny RIcciardi:
I think they're asking a lot of, as it relates to ESG, it's carbon emissions, fossil fuels. A lot of the diversity, so the S in ESG.
Sar Ruddenklau:
Keeping on the human and the diversity track, there's this common fear that AI is going to replace us all. It's going to replace all our jobs. What's your take on this? How can the role of humans in the technology loop continue to evolve, to actually maximize the usefulness and the utility of things like AI?
Joe Capobianco:
I think for the time being, AI is really just to make your life, your job more meaningful. You can focus on the more quality aspects or the more interesting items of your job, and maybe let AI handle a bit more of the administrative or boring tasks, if you will. I don't think we're at a point now where we're going to see jobs just wholesale drop off due to AI. I think there was a chart where since the release of ChatGPT, there hasn't been any type of meaningful driver in terms of AI impact to the workforce. For the time being, I think it's more of that assistant, Copilot type thing where we're not going to look at jobs just evaporating tomorrow.
Danny RIcciardi:
I agree with those sentiments. I think people need to embrace it and not view it as something that's coming after their job. There will always be something else to do.
I'm sure we all have a list of things that just don't get touched. To the extent you could use AI to cross off some of those easy, mundane tasks, it's just going to free up time to do things that are more a value-add to your organization.
Sar Ruddenklau:
There was a newspaper clipping floating around that we brought up in our prep call. I think it was a quote from an artist or an author who said, "I don't want AI to do my creative work, so I've got time to do my laundry and my dishes. I want AI to do my laundry and my dishes so I have time to do my creative work." That to me sums up what I want from AI in a nutshell.
I'm going to get you guys both to look deeper into your crystal ball. What do you believe is going to be the overarching impact of tech and AI on the whole trajectory of private equity? You can be as futurist as you want here.
Danny RIcciardi:
I think that technology will further streamline efficiencies and firms like myself will partner with, have more partnerships with firms like Passthrough in terms of their overall tech stack and use these new players that are coming with these great technologies to help drive efficiencies to the firm. Really, it's just evolving relationships within our organization. As time goes on, we find ways to use outside parties to help us with our current workflows.
Joe Capobianco:
I agree. I touched on it a bit in terms of the LLM, but I think where we're going with this is probably being able to query a dataset, produce a visual. When you think about the formatting it takes to do an AGM deck and all the visuals or different ways that you want to cut data. Can we have an LLM that's advanced enough to be able to produce a visual based on the inputs I'm giving it? I think that would be super cool. But we're definitely a ways off in terms of being able to implement just based on the accuracy that the models are returning and the fact that LLMs don't do math very well. In finance and private equity, that's not a great sign.
Sar Ruddenklau:
Investors are at risk of getting some really weird AGM decks is what you're saying?
Joe Capobianco:
That's right. That's right.
Sar Ruddenklau:
Well, thank you both for your time and thank you, Cassie, for helping me facilitate this conversation. I'm sure we can come back and revisit this in six months or a year, and all our predictions have gone out the window. That the whole AI landscape looks completely different again. Thank you both again, and thanks everyone for joining us.
Thinking of adding a digital sub doc solution to your fund tech? Talk to one of our experts today.
Your Privacy Choices