Across private equity, the AI conversation has moved from whether to adopt the technology to how to make it pay off across a portfolio. Hold periods are longer than ever, exits remain selective, and limited partners are pressing for realized returns. That combination puts more of the performance burden on the operating agenda, and AI now runs through nearly every discussion of how to deliver it.
That pressure shaped a recent gathering of operating partners, portfolio company executives, and functional leaders in London. Held under the Chatham House Rule, the conversations spanned AI, technology, talent, and operational execution, and the same themes surfaced regardless of which function was in the room. Many of them also echo what Owen Pagan, CEO and co-founder of Planr and a former operating partner at Investcorp, shared in a recent Private Capital Insiders episode.
Foundations Decide Whether AI Scales
Practitioners consistently framed the move beyond pilots as a question of foundations. Reliable data, clearly defined use cases, management sponsorship, and early attention to governance and security determine whether an initiative spreads across a business or stays confined to one team. The choice of model or vendor rarely explains the difference.
Owen Pagan describes the data requirement as a “steel thread” running from the portfolio company up to the board. The idea came from his time as an operating partner, when he watched a budget presentation he doubted could be delivered and realized he lacked the data to challenge it with anything more than instinct. He traces the weakness of traditional portfolio reporting to three issues: latency, consistency, and direction. Boards often review fourth-quarter results in February, after the window to act has closed. Priorities that dominate for two quarters fade before anyone tracks them consistently. And most reporting describes what already happened rather than what the data suggests is coming.
Waiting for perfect data is not a viable plan either. As Pagan notes, incomplete CRM records are a finding in their own right, because they expose a process gap that needs fixing. Collecting data across a portfolio also cannot depend on forcing every company into a single rigid template, which breaks down at scale. Frank Scarpelli put it plainly during the conversation: “If you don't have your data in the right shape and you don't have your processes in place, there's no AI silver bullet.”
That view is echoed in a recent KPMG interview, in which the firm's advisors describe fragmented and unstructured data as a core bottleneck and recommend a portfolio-wide AI playbook covering data standards, reporting expectations, preferred tools, and governance.
Tying AI to Outcomes the Board Recognizes
AI is reshaping work across finance, technology, commercial, talent, and operating teams. The more difficult task is connecting that work to results that show up in revenue, margin, or cash, and that a buyer will credit at exit. Many of the London discussions centered on the gap between visible AI usage and demonstrable business impact.
Pagan offered a useful frame for that gap, contrasting the show pony with the war horse. The show pony looks impressive in the boardroom: more AI-generated presentations, more polished material, more noise. The war horse is less elegant and far more useful, applying data and models to the operating questions that determine whether a plan gets delivered.
His example of the second category was specific. In one portfolio company selling into the education market, Planr's models flagged that the time deals spent at a particular pipeline stage had risen roughly 300 percent, something the business had never seen before. When the team investigated, they found a competitor had launched a new product and contacted the entire customer base, prompting buyers to pause and compare. The sales leader's forecast still assumed historical conversion rates. Because the signal surfaced early, the board could ask sharper questions and act before the quarter was missed, rather than learning about the shortfall two months after it happened.
The broader market is applying the same discipline. In the KPMG interview, advisors describe a move away from running pilots mainly to show activity to the board, toward wiring business targets directly into value creation plans and tracking hard financial metrics. They estimate that reworking high-frequency, programmatic processes could improve EBITA by 5 to 15 percent, while cautioning that some problems are better solved through process redesign or management discipline than through an agentic solution.
Execution Runs on Coordination and Alignment
Technology-enabled transformation depends on disciplined execution across leadership, operations, talent, and finance. Durable value creation begins with a short list of clear priorities, agreed with management and supported by the operating and sector expertise a sponsor can bring. Without that alignment, even well-funded initiatives tend to drift.
That alignment depends on how the sponsor engages. Pagan acknowledged that some portfolio companies push back on new data and reporting requirements, and that the reasons are often legitimate, from timing to workload. Where adoption works, sponsor and management end up looking at the same data set, which makes the relationship more collaborative. Teams spot problems together and address them together, and management retains control over how and when it shares its data.
A recent Fortium Partners analysis of the first 100 days of AI value creation makes a related case: standardize the decision process at the fund level, let each portfolio company own its priorities and results, baseline the metric before changing the workflow, and name one accountable executive for each outcome.
Talent Remains the Multiplier
Leadership quality, organizational design, and talent strategy remain among the most important drivers of enterprise value, and AI raises the stakes on each. The human capital operating partner role looks very different from one firm to the next. Participants shared a view that AI plans succeed or stall based on the capability and conviction of the leaders asked to carry them.
Inside portfolio companies, employees are asking direct questions about how AI will change their roles and what happens to the time it frees up. The KPMG advisors suggest turning the people closest to an existing process into product owners and change champions for the new one, which treats adoption as a leadership task.
Where the Conversation Goes Next
AI will keep raising expectations for the speed and precision of portfolio decisions. The discussions in London pointed to where the advantage will come from: consistent data flowing from every company, priorities agreed with management, an accountable owner for each outcome, and leaders who can bring their teams along. That groundwork is rarely visible in a board presentation, and it compounds across a hold period.
.jpg)