Full episode transcript: AI and PE: The Future of Value Creation

Guest: Giacomo Sonnino, Advisory Director, Charlesbank | Host: Kyle Roemer

Giacomo Sonnino: I don’t think there is an IC meeting right now where we don’t speak about AI. Everybody is looking at this lens, and whenever you look at the business, you start thinking about “what is the long-term value or the terminal value of this business that we would be considering for an investment?” And I would guarantee that every single IC is going through it, for every single investment.

Kyle Roemer: Welcome back to “AI and PE: The Future of Value Creation.” We’re thrilled to have Giacomo Sonnino from Charlesbank with us today. As Charlesbank’s Advisory Director, Giacomo drives operational improvements and value creation throughout their portfolio companies. Prior to joining Charlesbank, Giacomo served as the North American COO of QuantumBlack, McKinsey’s AI and analytics consulting arm. Welcome, Giacomo.

Giacomo Sonnino: Thank you, Kyle. Thank you for having me here.

Kyle Roemer: Yeah, thanks for joining us. Look, we’re really excited about today’s conversation. It’s going to be focused on exit readiness and what “AI ready” actually looks like for PE-backed businesses. So Giacomo, I thought we’d start there. When you think about AI ready at exit, what does that actually mean in practice?

Giacomo Sonnino: You can create value with AI today, so you better be creating value with AI at exit. This answer is probably very different than the answer I would have given you 12 months ago. Twelve months ago I probably would have said, look, you need to have an AI roadmap. You need to have some pilots, ideally some proof points. Ideally, think about how your offering will change with AI, if relevant.

That is probably not sufficient anymore. Today you need to have a clear articulation of how AI is going to affect your business model, how you are addressing AI with a clear roadmap linked to your value creation plan. You have to have some proven use cases, some real AI implemented in your workflows, some demonstrable benefit, and ideally an offering that has already changed with AI.

That’s probably what AI exit ready means today. Twelve months from now, obviously, it will be different.

Kyle Roemer: It’s so interesting, the point on 12 months ago versus today versus 12 months from now. It’s moved so incredibly fast. With that pace, with all the change, how are you distinguishing between companies that are differentiated versus ones that are just keeping up?

Giacomo Sonnino: Companies can create value with AI in three ways, at different levels of sophistication, obviously.

Number one is general productivity. You and I and anybody else can use Claude, can use some meeting productivity AI tool, email productivity, and so on and so forth. That’s helpful. It gives us hours back, it may give us better output. Do we see bottom line from that? Maybe, but not directly.

Then there is what I call the proven use cases. And there are a lot of proven use cases now that you can just go and execute, where there are either third-party solutions or it’s really known how to do it. I’ll give you a few examples: software development productivity, sales and marketing effectiveness, customer experience, and what I would broadly call the back office: finance, legal, and so on and so forth. These things are proven. There are solutions out there that are ready to implement. People know how to rethink workflows. I mean, you guys know very well how to do that in the finance space. And the risk-reward on this is pretty clear: low risk of implementation because they’re proven and known, and speed to impact is also generally quite fast.

And then the third, a more exciting way, is what I call AI in the core of the business, which actually means very different things by business. If you are a software company: is your product AI, does it include some AI enablement? If you are a business like a wealth manager, the core of your business is your advisor. Are you driving advisor productivity with AI, or lead generation with AI? That’s the core of your business, and so on and so forth. That’s a lot harder. You actually have to rethink how work is done. You have to redesign your workflows, you have to redesign your tech stack, and you actually have to drive a lot of change management.

Now, you ask what companies are differentiated. Those that are differentiated have built a system to create value with AI that tackles these three potential ways of value creation, are increasingly going into the third one in a very systematic way, and have built an infrastructure around it.

Kyle Roemer: You’re part of a lot of potential deals and businesses that you evaluate. Are you seeing AI more explicitly come up in buyer conversations and diligence processes today for deals you all look at?

Giacomo Sonnino: I don’t think there is an IC meeting right now where we don’t speak about AI. Either at Charlesbank, but I would imagine that is pretty much every IC in the PE industry.

Everybody is looking at this lens, and whenever you look at the business, you start thinking about how this business will operate in, I don’t know if I’d call it “a post-AI world,” but definitely in a few years, given AI. What is the long-term value or the terminal value of this business that we would be considering for an investment?

And I would guarantee that every single IC is going through it for every single investment.

Kyle Roemer: Companies are going through their AI journeys at various stages of adoption. Are you finding there are common areas where companies are getting stuck in accelerating adoption, moving from pilots to production? What are you finding in the market right now?

Giacomo Sonnino: The most likely failure that I observe, the failure point or failure mode, is a company gets stuck in pilot purgatory. Companies keep doing one, two, three, four, five, ten, twelve pilots as proof points and never get out of that. I actually try to avoid pilots at all costs. But the common mistakes that get to that point, I’ll give you four failure points specifically:

Number one, you see the AI program led by the CTO or CIO. So you take it as a technology project, not as a business problem. Very little CEO involvement, or no involvement, or in general no business leadership involvement. And then not thinking about how this is a business project: the impact, the change, the workflow, how we change the way we do the work.

These common mistakes lead you to doing small trials that never get traction in the business, and that get stuck into what I define as pilot purgatory, which is: you have some hints of “this could be interesting,” but you don’t have the buy-in to drive the large-scale transformation to actually change how the company does the job.

I can give you the counter to this, which is: what do the companies that actually get it right do? I articulate it into three things.

Number one, you need to be AI ready. What does that mean? Well, you need to have a CEO that has high conviction and vision on this. You need to have an AI leader in general, someone appointed at the C-suite level. Not a technical person, but a businessperson that can lead this initiative. And you need to have a clear AI strategy or roadmap of how your business is going to change through it.

In addition to it, you need to have a clear understanding of what the impact of AI is on the revenue and margin of the company, at a granular level. Not at the overall company level, but business by business, product by product, offering by offering. I would say both short term and long term, because that is likely going to be different.

And then you need to get going with a clear infrastructure to execute: the technology build, the workflow changes, and change management. Without these three things, you’re not going to get any bottom line impact of any meaningful size.

Kyle Roemer: Yeah, I like that intersection there. It’s one thing, to your point, that having a CIO or CTO lead these initiatives is not necessarily the right starting point. You mentioned getting the CFOs or CEOs fully engaged, fully bought in, appointing someone on the AI leadership side, having a roadmap, having the infrastructure, etc. There’s a lot of things that you have to do there and get right.

Do you find it’s the definition of the roadmap? The CEO buy-in? Is it the resourcing side? Where are people finding more obstacles navigating today in the portfolio?

Giacomo Sonnino: In general, the majority of CEOs would say “well, I’d love my team to do it,” which is great. You want your team to own it. But as you and I know very well, the skill sets needed to do this technology building and change management most likely do not exist in the majority of companies. So you move to the next step, which is: okay, how do I create the environment for success? Which means, can I go hire and find the right talent? Like an AI transformation manager, or AI talent that can actually go build technology, serious project managers that can drive this technology build or product changes, an infrastructure to do training, and so on and so forth.

And then very frequently you find that hiring that talent and integrating it into the organization is really hard, and the skill gap that you need to fill is really big. So what do you do? You default to third parties. You go out to the market to find third parties today and you find a lot of marketing. And it’s very difficult to navigate what’s real and what’s not real. People just seem to make big statements, and figuring out what’s actually delivering and who can deliver versus who cannot is really hard.

It’s very hard even for me, and I’ve been doing this for ten years. So someone new will find it a lot harder.

Kyle Roemer: Yeah, I think you could throw a rock and hit thousands of consulting companies that say they do AI work.

Let’s talk about an example from your portfolio. You’ve got a large portfolio of companies today. Tell us about a leader in your portfolio from an AI perspective. What are they doing? What makes them a leader today?

Giacomo Sonnino: I’ll give you four examples, two in our business services sector and two in our technology and tech infrastructure sector. They all share very common features that lead them to success. Namely, that would be Accordion and Aprio on the business services side, and Symplr and EverDriven on the technology side. What do these companies’ CEOs do?

Number one, extremely high aspiration on what AI could mean for the business. Very clear leadership.

Number two, a very, very clear roadmap: How are we tackling the business operations? How are we tackling our offering? How are we effectively using AI in support of the value creation plan? I can give you exactly how they articulated it one by one. There is no doubt about how AI contributes to their business.

Number three, they put an enormous amount of resources against this topic because they realize that it is absolutely fundamental for them.

And number four, very rigorous governance and reporting, both at the company level but also at the board level. This is clearly a board topic for all of the companies I mentioned.

These are business leaders that realize that AI is fundamental for them and they’re playing a very offensive role in these transformations.

Kyle Roemer: Yeah, it’s helpful to hear the examples. I think that resonates with folks. If I’m a services business or I’m a technology business, I can get my head around, or my arms around, what this might look like. In your role, you’re in a unique position on the advisory side, the operating side.

How do you think about partnering with portfolio companies on embedding AI, but doing it effectively to have the biggest impact? What are some of the things that you and your team do?

Giacomo Sonnino: As you know, the operating partner job is a little bit of a ninja job. It requires you to deploy very different skill sets and deploy them in very different ways depending on the needs of the company.

I’ll give you two different operating models in which I see ourselves operating. Actually, I’ll say three.

Number one, part of my job is to work with the management team to raise the aspiration and help put in place the conditions that I’ve just described. Aspiration from a CEO perspective. Do we have a leader in place to drive this? No? Let’s go hire him or her. Do we have a clear AI roadmap linked to the value creation plan? Do we have resources that can go and execute? Call it truly a counseling and advisory point of view.

Then there’s a second step, which is: look, there are only so many operating partners in the world and there are a lot more companies, so you need third parties to do the work. I spend a lot of time curating a network of vendors and external advisors that we can deploy in the portfolio that have different levels of skill sets and different levels of sophistication, so I can help our portfolio companies select the right vendor for them, or for what they’re trying to do, and accelerate the deployments.

And number three, on a case-by-case basis, when the ROI is extremely high, myself and some of my peers that focus on AI can deploy into portfolio companies to help accelerate some of these programs.

There is a trade-off at any point in time of how you optimize your deployment versus ROI, because when you go do these AI transformations, there is only so much a human can do. You can do one, maybe two, but you certainly cannot do seven or ten of them. You need a network of vendors to support that. And ideally you want management to own it and drive it. So it’s not even the right answer for an operating partner to go do it in every single company in the portfolio. That’s not success.

Kyle Roemer: If you think about companies you’ve looked at, and companies even in your portfolio, is there a side of that story that tends to resonate more with buyers? Or is it a combination of both internal and external? What are you seeing right now?

Giacomo Sonnino: It depends, is the answer. But it depends on two dimensions. Number one, it depends on the business type, and it depends on the size of the prize. What brings it all together is that you have to do a prioritization of AI against the value creation plan.

I find it hard to imagine that you’re going to market a transaction knowing that, let’s say, you could serve clients in an improved way with AI, and not having done it or not having a plan to do it.

So if your front office can be disrupted with AI, you should probably think about how you address that point. At the same time, you can get tremendous efficiency, effectiveness, and margin improvement with AI. You should do that too.

The question becomes, what is the balance? And it all depends on the size of the prize. If you need to improve your margin, there is plenty of margin to go after with AI. I’m thinking about software development productivity, thinking about how you can reimagine the finance function with AI. How can you reimagine your legal function with AI? How can you reimagine your back office processing with AI? There is plenty of that. You can go there. Whether you go do that versus the front office depends on the size of the prize and what the priorities are.

Kyle Roemer: Yeah, and to your point, it’s a balancing act. But the size of the prize is really important. If the prize is big enough within a reasonable time period, you should probably go after a number of things. So that brings me to this: we talked about what’s resonated with buyers, but there’s this term called “AI theater” that happens in diligence and through a process. First, explain AI theater. But second, how are buyers like yourself seeing through that theater?

Giacomo Sonnino: A lot of companies think about, call it, what they can dress up for an exit process. And I would say people are naive if they think they can dress up with an AI or training roadmap. It’s fairly clear to see if you’re serious about AI or not, and if you’re seeing value from AI or not.

For example, everybody, as I mentioned, that is going to market in the next, let’s say, 0 to 12 months will have an AI roadmap in their CIM. Great. There will be companies that have reinvented their products with AI, or products and services with AI, and companies that have not. Clear demarcation. Have you changed your offering? There are companies that have completely redesigned their processes. There are companies that are going to say that they may redesign their processes. And there are companies that say, “we’re fine, we’re not going to do that because AI doesn’t touch us.” It’s actually quite binary to see where you are.

The only gray area, I would say, is when people are starting early, doing some pilots, and not actually having fully done the transformation. What you would typically find is people make a statement like, “oh, we have demonstrated X many thousands of hours of efficiencies.” Great. But what is the bottom line impact of those hours? That answer is where the rubber hits the road. In many cases people will say, well, that creates capacity in the system, but we haven’t yet seen the uplift in the system for those hours, and we have not seen the cost reduction from that, because we are kind of stuck in between. I call it, again, going back to the concept from earlier, pilot purgatory.

That is the moment of gray area where you need to understand: okay, are we really going to go through with these savings and these initiatives, or this uplift that we can get because we have more productivity, or not? Everything else, I would say, is pretty binary.

Kyle Roemer: It’s such an interesting term, theater. I think back to the idea of moving from pilots to production. That’s always been around in the data science field, and that was always a measurement we looked at: are we moving from pilots to actually putting these things into production, into a workflow, into a process, a system? I think that’s even more true with AI.

It’s one thing to have individuals in a department or function using the tools and being more efficient. It’s another thing reimagining that workflow with AI so that you might need the same amount of people, you might need less, they’re significantly more efficient, and you’re actually seeing a quantifiable benefit versus what can oftentimes feel like a cost avoidance type of narrative versus real hard dollars.

Giacomo Sonnino: For sure. I completely agree with you. And the reality is, look, I’m making it oversimplified to an extent, but my recommendation to almost everybody is: whatever company you’re looking at, in whatever sector, find someone that intersects from an expertise perspective between AI and that specific sector. You’re looking at a wealth manager? Find someone that knows AI in wealth management. You’re looking at whatever software company? Find someone that can go diligence it, someone that knows AI in software development.

And you have to go extremely, extremely granular. The high-level statement of what AI can do to software development is just not going to be helpful in underwriting.

Kyle Roemer: That’s right. Yeah, that’s right. It goes back to the notion that the deeper you are in a workflow, in a subsector, in a subsegment, the more real it becomes, and I think the more value you can create with it.

Well, look, we haven’t really touched on finance yet, so I think it’s about time we do. We spend a lot of time here in finance and back office, and we see AI drive a ton of really near-term and intermediate ROI. From your view, where are you seeing AI today within finance organizations having the most impact? And maybe juxtapose that with what that might have looked like 6 to 12 months ago, because it’s moved so significantly in recent months.

Giacomo Sonnino: Let’s start with the title here. Finance is a proven opportunity. AI can transform the finance function. There is no doubt. There are so many use cases. You have to do it. That’s the title for me.

There are two subtitles to that: What can you do today as a CFO of an existing company, and what would you do if you were to build a finance function from scratch, or what will the finance function of the future be? Those are two different worlds that you have to think about.

Let’s start from what you can do today. There are technologies, solutions, and solution providers that can help you drastically change how the finance function works, how the people in finance spend time, and the value that you get out of the finance function.

In the past, and in many companies, we see a lot of resources in finance doing, I call it, important but boring transactional stuff that can actually be done a lot more efficiently and effectively with AI, and free up a lot of time and a lot of money for finance to be a real thought partner to the business.

I’ll give you a practical example. Every single portfolio company in the PE space has to do a 13-week cash flow. Every single one. There is no company in the PE world that doesn’t do a 13-week cash flow.

The enormous majority of these companies, in the past, have done it through an Excel that gets run by someone that spends 50 to 100% of their time running that process: collecting files, consolidating AP and AR, connecting bank information, etc. In today’s world, there is no reason, literally zero, for which that process needs to be run by a full-time person or 50% of a full-time person.

You can rethink your process with the help of Accordion and others. You can use technologies that are out there that actually give you a live cash flow on a day-by-day basis by connecting your ERP, your bank accounts, your general ledger, etc. There is no need for a person to be involved beyond looking at the result and thinking about the liquidity of the company, not spending the time processing it.

There are plenty of these examples. Close: how many companies spend 15, 20, 25, 30 days closing? There are plenty of technologies that help you do that. Think about FP&A. There is plenty of AI out there that can help you optimize your FP&A processes, improve your accuracy on forecasting, improve your efficiency in producing the forecast and analyzing the variances, and so on and so forth.

Those are just in the core of finance. But the reality is you want to use those technologies to free up costs, free up capacity, get better outcomes by the way, and then use that time and those resources to go do things that are a lot more impactful for the business.

Think about RCM, think about pricing analytics, think about cost analytics. The reality is the finance team might not even have time to do those today, but those will actually be a lot more valuable in shifting and creating value for the business.

I would think about finance today as a land full of opportunities. Go process by process: how can you rethink the workflow? I’m not suggesting that every CFO goes and does 20 of them at the same time, but I do think that CFOs need to think about how they’re going to transform what they do, and start with the thing that is most burning and most important for them. It could be close, it could be cash flow, it could be planning, it could be analytics. But you need to have an answer.

And then the most visionary CFOs and the most visionary companies would actively be thinking about the shape of the finance function three years from now, four years from now, what I would call the finance function of the future. You certainly would not start with the tech stack you have today and with the org design that you have today.

So you actually have to think about what you want the finance function to be, and do it very differently, because you can have a lot more value with a lot less cost.

Kyle Roemer: Yeah, that’s right. And if you’re reimagining what finance looks like in the future, you’re not always beholden to the legacy of how you did things over the last couple of years, even with some efficiency gains with AI. Imagine systems talking to each other with agents. Imagine agents doing a lot of these activities, and folks not having to pull data, reconcile data, map data in Excel to PowerPoint, to all these various disconnected tools.

There’s a real future where these teams are much smaller, but even more impactful, with agents and reimagining these functions with AI at the forefront versus tacking AI onto existing processes.

Giacomo Sonnino: For sure. And I would even add, in the past we were limited by the available technologies. There were only so many, and they were costly, difficult to implement, and fairly rigid. Thoughtful companies and thoughtful CFOs can think about: what are the 2 or 3 agents that can drastically change what I do and let me create more value?

In the past, you needed enormous teams and enormous cost to go do that. Today, the right team can do it fairly quickly. So you can start thinking about doing very sophisticated things: let’s figure out what the three or four anchor systems are that we want to use; let’s figure out what agents we need to build to operate or connect those systems; and let’s think about how the humans interface with this world of systems and agents through what I call an orchestration layer, a simple platform on top of everything else that helps us be more efficient, more precise, more productive. Building something like that 2 or 3 years ago would have cost millions of dollars. Today, it is actually a lot more affordable and more feasible given the tools that we have available.

And I also find that the technology innovation is moving so fast that I think probably six months from now, you and I would be having a very different conversation on what’s possible with AI than what it is today, and certainly it’s different than what we thought six months ago. All of this, in terms of the finance function of the future, finance use cases, agents in the finance function, it’s just very real.

Kyle Roemer: Yeah, I completely agree. We’re seeing it now: you can move much faster, you can build much faster with smaller teams. The one thing I want to round out this part of the conversation with is: all of that is great, but the change management, the transformation that has to occur within these organizations, how do you think about going through implementations without overwhelming the entire organization?

Giacomo Sonnino: I’ll leave you with two takeaways. Number one, if you want to be successful, you have to put real time and effort into change management.

And number two, it’s fairly unrealistic for a mid-market, PE-owned company to do more than one, two, or three of these a year. You’re certainly not going to go to ten. You may do one or two, maybe three. There is just not enough capacity in the system to absorb all that change management.

Therefore, connecting to what we said earlier in the conversation, you need to know your AI roadmap. You need to know where your AI roadmap links to your value creation plan. You need to know what ROI you’re expecting, and you have to go after that ROI. You’re going to go after the big things, and you’re not going to do ten of them because you don’t have the capacity to drive the change management. There is no organization that I see today that is doing ten use cases at scale in a single year. Not going to happen.

Kyle Roemer: Yeah. Something you said resonates so much with me. The change management piece is just significant across these businesses, and I think under-invested in a lot of the time.

I want to wrap with maybe one question for you, Giacomo. For portfolio companies, maybe earlier in their hold periods or their life cycle, what would you recommend as the single most important thing they can do right now related to being exit ready with AI and AI transformation? What do you think it is?

Giacomo Sonnino: You have to think about how your business is going to look in a post-AI world, and I define that by: if something can be automated or redesigned or redone with an AI-enabled workflow, someone will do it.

In other words, if it can be automated, someone will automate it. So better you do it to your business than someone else does it for you. But you have to do that thinking.

Kyle Roemer: Giacomo, thanks for joining us. This was an awesome conversation about what it actually looks like to be AI ready for exit. Some of the insights around CEO engagement, having folks dedicated to AI transformation with clear, quantifiable benefits, I think it rang true that you’ve got to start now and, frankly, be ready for a lot of the change management that’s going to be required coming out of this.

Whether it’s internal use cases, external use cases, the opportunity in finance, it was just a really rich conversation and I appreciate you being part of it.

Giacomo Sonnino: Thank you.