Two weeks ago, operating partners, value creation leads, and portfolio company executives from PE and VC firms spent a day together in San Francisco at the Private Capital Global West Coast Value Creation Exchange. The day was built for cross-functional discussion, and by the end, the AI, talent, sales, marketing, and finance leaders in the room kept arriving at the same problem from different seats at the table.
When AI adoption stalls inside a portfolio company, the cause is almost always organizational. No one owns getting the tool into daily use. The economics behind a piece of software shift, and nobody updates the model. Talent partners get pulled in after a leadership gap has already slowed the plan. The CFO lacks a clear mandate to hold every function to the same version of the plan. What follows looks at where each breakdown happens and what the firms avoiding them do differently.
Getting AI to Work Across the Portfolio
The hard part of AI implementation today is getting a team to use it consistently and then carrying it beyond the portfolio company where it started. A pilot can show early promise and still go nowhere. Its impact depends on whether it fits how employees already work, solves a problem they recognize, and produces results the business can measure.
Anyone who lived through the ERP rollouts of the early 2000s has seen this pattern. The software went live on schedule, and the value showed up years later, once someone owned the process changes around it. Value creation teams can shorten that lag by working with portfolio leadership to pick applications that matter, assign clear ownership for adoption, and fix the data and systems problems that block it. A business on legacy systems and fragmented data needs a different first move than one with clean infrastructure.
The firms in the room at the PCG event described AI adoption as an ongoing discipline. They revisit which use cases deliver results, retire the ones that don't, and carry what works into the next portfolio company so each evaluation builds on the last.
The New Economics of AI-Enabled Software
The economics of AI-enabled software run counter to the assumptions most underwriting models were built on. Classic software investing rested on one mechanic: build the product once, and the marginal cost of serving another customer approaches zero. AI breaks that mechanic. Every inference call carries a real, variable cost that keeps scaling with usage long after the engineering work is done. Forbes reported in late July that the economics differ materially by business model: OpenAI was estimated to operate at roughly a 33% gross margin, weighed down by flat-rate consumer ChatGPT subscriptions, while Anthropic—whose revenue is predominantly enterprise-driven—reportedly lifted its inference margin from 38% to more than 70% during 2026.
Diligence now has to model token and infrastructure spend with the same rigor it applies to customer acquisition cost or churn: tracked, stress-tested, and built into the base case. The shift also sharpens the defensibility question. A feature built on someone else's model is easy to replicate and easy to commoditize the moment a larger platform decides to absorb it. Durable differentiation increasingly comes from proprietary data, workflow ownership, and deep infrastructure connections that make switching expensive for a customer even after the underlying model becomes a commodity.
Talent Partners Across the Investment Lifecycle
Talent partners still find the right leaders, and their involvement now starts well before a role opens. In diligence, they assess the leadership team against the investment plan and flag gaps that could affect execution. After close, they help shape the team and support integration.
That earlier involvement changes what the job requires. Assessing a management team in diligence means judging whether the people already in place can execute the specific plan the deal team is underwriting, and identifying early where a hire, a restructure, or additional support will be needed. A talent team that assesses against the plan itself functions as a genuine extension of the deal team.
How AI Is Changing Hiring and Workforce Planning
Current AI tools help talent teams work through candidate data, surface potential matches, and spot gaps across a leadership team far faster than a manual search. Firms also need to rethink what they screen for in new hires and existing teams: people who understand where the technology helps, are willing to learn, and can adapt as roles change. For talent partners, the question now covers the full hold period: who to hire today, and how to build a workforce that can carry out the value creation plan as the tools keep changing around it.
"The tools available to talent partners have finally caught up to what the job requires," said Frank Scarpelli, Managing Partner of Sparc Partners. "I've spent time with the team at Findem, and their work on signal-based candidate data is a good example of where this is heading. The tool surfaces patterns in someone's career that predict whether they can execute in a specific environment. That gives the talent partner a far better starting point and leaves the judgment call with them. Every mandate now carries an AI-fluency component, so the firms and talent partners who adopt these tools early will have a real advantage."
The CFO's Role in Executing the Value Creation Plan
A value creation plan only works if the portfolio company's leaders can translate it into decisions across the business, and that includes the CFO. Working alongside the CEO and other functional leaders, the CFO sets financial priorities, tests whether initiatives deliver expected results, and keeps teams focused on shared goals: scaling the business, improving margins, reducing costs, and growing revenue.
The role matters most when priorities compete. A new investment in sales, a pricing change, and an effort to simplify operations may all support growth, and each carries different costs, timing, and risk. The CFO helps the leadership team weigh those tradeoffs and see how a decision in one function ripples through the rest of the plan.
Better Tools for Finance Teams
AI is also making it easier to organize data, build dashboards, and track metrics across a portfolio. Leaders can spot performance changes as they happen and dig into what is driving them, well before a quarterly report surfaces it. Finance teams still have to choose metrics that reflect the value creation plan, confirm the underlying data is reliable, and bring the right people into the conversation. Used well, these tools show the leadership team what is working, where execution is slipping, and where to focus next.
Key Takeaways for Those Not in the Room
The thread running through the day was coordination. A pilot that never scales, a software deal underwritten on outdated margin assumptions, a leadership gap that surfaces after close, a hiring process that screens for yesterday's skills. Each one seems to belong to a different function, and all four trace back to the same root cause: a value creation team that isn't coordinating early enough or closely enough to catch them.
The practical starting point is to make every part of the value creation plan someone's responsibility. Decide who evaluates which AI use cases deserve to scale across the portfolio and who shuts down the ones that are not delivering. In software diligence, model token and infrastructure costs alongside customer acquisition costs. Bring talent partners into underwriting early enough to identify the leaders the plan will need. Give the CFO the mandate and the tools to track progress against shared goals for growth, margins, and cost.
AI adoption is table stakes. The results go to the firms that start with the business case.
