Building AI Capability in Private Equity: Fund Level, Portfolio Company Level, or Both?

As private equity firms move beyond AI experimentation, a foundational question is emerging: Where should AI capability be built and owned? The answer shapes talent strategy, governance overhead, implementation speed, and ultimately what value travels with a portfolio company at exit. 

Key Takeaways:

• Before selecting AI tools or use cases, PE firms need to decide whether AI capability should live at the fund level, the portfolio company level, or a hybrid of both. 
• This is not just a technology question. It is an investment thesis question that affects how value is created and retained. 
• Talent scarcity, governance consistency, and hold-period economics all factor into the decision. 
• There is no universal right answer, but firms that address this question deliberately will move faster and could avoid costly corrections in the future.  

Why This Decision Comes First 

While most AI conversations in private equity jump quickly to tools and use cases, including processes to automate, vendors to evaluate, and how ROI is measured, the more important question is where each AI capability should live, whether at the fund, portfolio company, or both. If a firm has not decided the answer to these questions, then each portfolio company ends up solving the same problems independently. Hence, governance becomes inconsistent, and funds miss the opportunity to build institutional knowledge that compounds across all their portfolio companies. 

Further, the two primary models, fund-level AI and portfolio company-level AI, carry different implications for how firms invest, operate, and eventually exit.  

People: Who Owns the Capability? 

Many middle market portfolio companies face challenges when it comes to hiring dedicated AI leadership due to the high cost, driven by a competitive market and high expectations.  

One solution is a centralized approach where all portfolio companies can share a resource: for example, a single experienced hire who works across multiple companies or a small central team. However, a shared resource does not eliminate the need for portco-level engagement, but it makes the AI process much more streamlined. 

Another consideration is the type of talent portfolio companies prioritize. Early-stage AI efforts typically benefit more from business-oriented leadership focused on change management, use case identification, and adoption rather than from technical developers building custom solutions. Technical depth becomes more valuable once leaders have validated where AI can create real impact. 

Change management in particular is important. Underinvestment in training and communication is one of the most common reasons AI initiatives underperform. Hence, a streamlined approach makes it easier to systematize change management through shared playbooks, training resources, and communities.  

Process: Who Sets the Playbook? 

The degree of standardization is contingent on both portfolio composition and operational thesis.  

Back-office functions such as accounts payable, financial reporting, and HR administration tend to look similar across portfolio companies regardless of the specific industry. Hence, these functions are suitable to adapt standardized approaches developed once and deployed many times, reaping significant efficiency gains.  

However, customer-facing and industry-specific workflows can differ. For example, a healthcare portfolio company and a manufacturing portfolio company have fundamentally different operational contexts and forcing standardization where differentiation matters creates friction rather than value. 

Governance is another area of consideration. Regardless of where implementation decisions are made, AI policies, vendor selection criteria, data-handling requirements, and risk thresholds should be consistent across the portfolio. Inconsistent governance creates inconsistent risk exposure and makes it harder to learn across companies. 

The same logic applies to measurement. Without a common foundation for tracking adoption, usage, and business impact, firms cannot compare results or identify what is working and what isn’t working. The specific metrics may vary by use case, but the measurement requires consistency to enable portfolio-level analysis. 
 

Technology: Who Makes the Platform Bet? 

The question of whether to standardize specific AI tools is often a technology decision, but it is really about which ones offer the best options for speed, flexibility, and leverage.  

Standardization for a core set of tools, particularly for foundational capabilities like productivity AI, creates procurement leverage, simplifies security review, and makes it easier to share learnings across the entire portfolio. It also reduces the burden on individual portfolio companies to evaluate and vet vendors independently. 

However, the trade-off is flexibility. Standardization works well for common requirements but can constrain portfolio companies who demand specialized requirements. Many firms ultimately come up with a hybrid approach: a portfolio-wide stance on foundational tools with flexibility for domain-specific needs. 

The build versus buy question follows a similar logic. Commercial tools can offer the best option for speed and dedicated vendor support. Firms might be able to consider custom development in the future, once repeatable high-value, differentiated workflows have been identified where commercial tools fall short and have the technical capability to build and maintain solutions over time. 

Making the Decision 

There is no universal solution to the architecture. The right approach depends on context, including portfolio composition, hold period, operational thesis, and the existing shared services infrastructure. 

Firms with portfolio companies that operate similarly and that employ a value creation thesis centered on operational improvement have more to gain from platform-level capabilities. On the other hand, those with diverse portfolios or a lighter-touch operational approach may find that portco-level flexibility serves them better, with governance as the primary area for standardization. 

The key is to take a proactive role in making the decision rather than letting it happen by default. Firms that address the architecture question early can build coherent strategies that compound over time. Firms that don’t address it often find themselves unwinding fragmented approaches later at significant cost. 

Looking Ahead 

Although whatever decision firms make for architecture is not permanent, it can be expensive to reverse. Capabilities built at the fund level create dependencies that portfolio companies may struggle to replicate at exit. Capabilities built at the portco level may duplicate effort and miss opportunities for shared learning. Getting this decision right does not guarantee AI success, but it sets the foundation for everything that follows. 

If you are evaluating how to structure AI capability across your portfolio, our team can help you assess the trade-offs and develop an approach that fits your investment thesis and operational model.