Innovation in traditional industries like consulting, law and finance is no task for the faint of heart. While most investors and financial professionals agree that change will eventually come to their function as AI progresses, to embrace it and push for it is another mindset entirely.

The ADKAR model for change management is one way that firms we have spoken with think about individual transformations within their portfolio companies. While enterprises have used this model for some time, some investors believe more funds should consider implementing it within their own organization to lay the groundwork for new technologies. What distinguishes ADKAR from other change management models, is a focus on creating awareness and excitement for change in the present and transitional states of the organization, so that the full team is more ready to participate in any new efforts.

How does a PE Vice President drive change?

The ADKAR model envisions involvement from the entire team, from general partners all the way down to analysts. If private equity associates are the front-line experts with the knowledge to develop ideas for transformation in the investment process, vice presidents are the visionaries with broad experience and management connections to propel changes over the line. Those who have been through it describe the VP role in private investment firms as one of the most demanding, but also the most rewarding in terms of the responsibilities and opportunities it offers.

For those unfamiliar, the role is often considered the most challenging because VPs are expected to perform strongly in everything from sourcing, to due diligence, advising on deal structuring, modeling, presenting in investment committee meetings, deal execution and company reporting - all while managing deal teams. From a promotion ladder perspective, an investing professional typically reaches the VP level upon advancing from an associate or senior associate level, so these investors typically have plenty of experience investing and analyzing deals.

The big shift in workload is often considered to be going from an individual contributor, deal-focused role to a more company-wide management style role, while still being involved in the front line work. For all of these reasons, VPs tend to be heavily involved in change management, setting the tone for company practices, both trickling down to management initiatives to ground level teams, and funneling up feedback based on what is seen in the market. Perhaps because it is a role that requires such a high level of personal investment, the VPs we spoke with seemed excited about the future of private equity, and their role in both executing and transforming prevailing investment models.

Influencing Change at all Levels

For this piece, Capital AI spoke with two private equity vice presidents at large, well known funds based in the US. As with last month’s post on the PE associate’s experience with technology, the approach taken by different professionals in the VP role can vary widely depending on their work styles and the firm’s priority. If we had to nail down one commonality, however, it is that we consistently saw VPs more passionate to think about the role of AI across various parts of the deal process and even within portfolio companies. They were also more optimistic about getting it implemented, even if they acknowledged that the firm was not quite ready.

One VP at a growth fund primarily focused on tech and SaaS, provided a compelling view of how AI is being leveraged across three main areas:

1. Investment in AI-driven Companies: Many fund leaders are considering how and at what stage to start investing in AI companies. Although this particular fund was cautious about investing directly in AI companies due to high valuations and a lack of in-house technical expertise, they recognized the potential impact of AI on technology investment theses.

“We don't do it yet (right now), and AI valuations are just through the roof. We also don't have the technical expertise to tell a good one from a bad one. Lastly, we need to build a network to help build out deals. So we don't invest in AI today, but we recognize AI will become a major part of SaaS portfolios going forward.”

2. Operational Efficiency within Portfolio Companies: A significant portion of AI integration occurs at the portfolio company level. For instance, a VP at a major technology investor detailed an initiative where a D2C brand implemented a customer service AI chatbot that successfully handled 85% of all inbound queries within three months, leading to a reduction in the customer service staff by 25%. Such implementations underscore AI's role in enhancing operational efficiency and cost-effectiveness.

“Within three months, it was handling 85% of all inbound queries, which led us to streamline the company's customer service department significantly,” he noted.

3. Enhancing Investment Processes: The use of AI tools like ChatGPT and Copilot, although proceeding cautiously within large funds, illustrates ongoing experiments within investment practices. These tools assist in initial deal assessments and industry analyses but have yet to replace the depth required for investment committee discussions. However, they serve as valuable starting points for further analysis.

One firm we spoke with had recently struck a deal with Microsoft to enable investors to access past deal information via an internal, on-prem LLM. "We use Copilot for preliminary deal assessments and industry analyses, though it’s not yet sophisticated enough for final investment decisions. It does provide a valuable starting point for deeper analysis."

One use case mentioned by several VPs was disseminating confidential investment memos (CIMs) to their deal teams, and asking analysts and associates to quickly flag relevant or unattractive deals based on summaries and pro-and-con lists provided by LLMs like Copilot’s. Importantly however, most investors emphasized they are not making any certain decisions based on the technology, let alone without human input.

Challenges and Promise of Implementation

The push-pull of internal versus external use cases for AI continues at many funds, with investors considering which technologies to test in house before deploying to a portfolio company, or testing more decentralized strategies.

Nonetheless, VPs excited about the promise of AI seemed undeterred about testing, even if LLMs did not yet have industry-specific use cases across the entire portfolio. Instead, many have looked function by function to identify opportunities that could work in nearly any vertical. For example, several mentioned tools like Jasper - relevant in marketing communications - or Harvey, a summarization and data extraction tool focused on legal workflows.

In fact, greater concerns arose about using more generalized tools like ChatGPT. “It's not thorough enough yet”, mentioned one VP we spoke with. It's not differentiated enough that I could just take what it gives me and take it to IC, but it's a good enough starting point. That said, I can look at a bunch of different companies across many industries and quickly get ramped up on key facts about that industry”.

Similar to the team CIM analysis effort undertaken by the technology investment firm, we talked to some team leaders who take the initiative into their own hands:

“What I would usually do when I get a CIM, I ask it, tell me the three to five key points of the investments and risks, and it will give me those, and then I'll read the sim and I'll keep those at the back of my mind. Usually, it will flag some of the obvious things, like, ‘revenue is not increasing, it's not profitable, but it's like the insights haven't been groundbreaking enough or like differentiated enough”.

Getting to the right tooling

Among everyone at the leadership level we spoke with, enterprise editions of Microsoft’s Copilot, OpenAI’s ChatGPT and Perplexity were among the most commonly used tools. Microsoft’s offering, in particular, came up frequently at larger funds as one that was at the “sweet spot in terms of keeping confidentiality and privacy concerns alongside analytical ability”.

Some VPs disagreed with their firm-level decisions, and continue to use consumer versions of ChatGPT on their devices to avoid the bias from past deals. Additionally, some felt the hype from their marketing and operations teams had surpassed the actual ability of the internal tooling.

“Now, people will start to sober up a little bit. And then once people sober up, it's where you actually see most of the benefits. People will sit down and spend some time to figure out how to use these tools beyond just chatbots, and in a way that is differentiated”, opined one VP at a major California PE firm.

Looking Toward the Horizon

Aside from what most investment leaders see as temporary challenges, many remain optimistic about the prospect of new use cases unlocked by AI. When pressed on whether the current hurdles would stop her from adopting, one VP replied “No, no - I'm definitely bullish on AI. I think in two to three years, there is no reason that it shouldn't be able to come up with 80 to 90% of some of the key points in a data room. The investment thesis and the risks, which would help a ton in terms of deciding if you move a deal forward or if you kill it, will be easier to identify”.

Another VP mentioned that the tedious work of disseminating and organizing information was at least “65 percent of the day”. During a deal process, aligning the buy- and sell-side teams, all of the bankers, consultants and other stakeholders, was a major theme that arose in conversation, with significant opportunity to be improved by AI and software in general.

Said one manager: “I’m still at a point where I probably review almost every cell in the Excel model, so figuring out how to efficiently review some of the technical analysis would be a major unlock.” Many technologists agree that incorporating LLMs into more quantitative exercises like modeling and calculations will be a major challenge, but if done correctly, will significantly increase both the trust and impact of the technology.

Investors with some familiarity of technology like mixed algorithms and RAG believe this will soon be possible. According to one East Coast-based VP, “the associates, 80% of the time, should be able to put the data into a template and then run a first pass at the analysis. There is no reason, in my mind, that you can't train a model to take the data, clean it (which currently takes lots of manual effort), and then run a first draft analysis.” Startups firms in the AI space, like Keye, agree the foundational technology is there, and are actively working on preliminary solutions.

“With a generic AI today, even if you prompt engineer it, is not quite going to be good enough. So we need to keep experimenting with ways of injecting context via RAG, mixed modeling, and other techniques until we crack this one”, according to a product expert at an LLM developer. In the meantime, it seems investment professionals will continue to wait with baited breath, or, better yet, actively engage with AI application developers to aim to be first to market with such a solution.

If you are among them, don’t hesitate to reach out to us at founders@keye.co to talk about what this could look like for your team.

‍