AI in the PE Value Creation Plan: The 100-Day Playbook

1. The Value Creation Plan Is Where Returns Are Made Now

For years, a lot of PE returns came from buying well and riding multiple expansion. That era has thinned out. With higher rates and full prices, more of the return has to come from making the business genuinely better. Operational value creation is no longer the nice-to-have on top of financial engineering; it is the main event.

That puts the value creation plan, and the first 100 days after close, at the center of the return. The VCP is where the operating partner lays out how this company will be worth more in five years: better commercial engine, lower cost to serve, stronger data, a more capable team.

AI is now part of that plan, because it touches most of the levers. But it works as a way to pull the existing levers harder, not as a separate "AI project" bolted on the side. This playbook is about building it in properly.

2. Where AI Fits in the VCP

The most common mistake is treating AI as its own workstream with its own goals. That produces pilots that demo well and change nothing. AI belongs as a layer across the value creation levers you already have.

On the commercial levers, AI improves pricing, sales productivity, and marketing efficiency.

On the cost levers, AI automates back-office work, customer support, and repetitive operational tasks.

On the data lever, AI is both the reason to fix the company's data and a beneficiary of fixing it.

The test for any proposed AI initiative is simple: which value creation lever does it move, and by how much in EBITDA terms? If you cannot answer that, it is a science experiment, not a value creation initiative. Tie AI to the lever, and the plan stays honest.

3. The First 100 Days: An AI-Aware Plan

The first 100 days set the trajectory. For AI, the goal in that window is not to deploy a lot; it is to see clearly and start two or three things that matter.

Notice what is not here: a firm-wide AI transformation. In 100 days you diagnose, fix the worst data gaps, and land one visible win that builds belief. The ambitious platform bets come later, on a foundation that exists. Trying to do everything at once is how AI initiatives stall before they show value.

4. The AI-Readiness Diagnostic

Before any AI initiative, you need an honest read on whether the company can absorb it. A fast diagnostic in the first month answers three questions.

Is the data usable? AI runs on data. If the company's data is fragmented, dirty, or trapped in disconnected systems, that is the first project, not the AI use case.

Where is the value? Which parts of this specific business have repetitive, high-volume work or pricing and commercial decisions that AI could improve? The answer differs by sector and by company.

Can the team adopt it? Capability and appetite vary enormously. A company with no analytics function needs a different plan from one with a data team.

This diagnostic is where our work usually starts, and you can get a first read with the AI Readiness Diagnostic. The point is to match the ambition to the reality, so the VCP's AI plan is grounded rather than aspirational.

5. The Commercial Levers

The commercial side is usually where AI creates the most EBITDA, because small percentage gains on revenue and margin are large in absolute terms.

Pricing. Often the single highest-return lever. AI helps analyze pricing data, identify where the company is leaving money on the table, and support more disciplined, segment-specific pricing. A point or two of price falls almost entirely to EBITDA.

Sales productivity. AI helps sales teams prioritize leads, prepare for calls, draft follow-ups, and spend more time selling. For a sales-driven business, modest productivity gains across the team compound.

Marketing efficiency. Content production, campaign analysis, and customer segmentation, doing more with the same marketing spend.

The discipline is to pick the lever that matters most for this business and pursue it specifically, not to sprinkle AI thinly across all three. For many portfolio companies, pricing alone justifies the whole AI effort in the VCP.

6. The Cost and Productivity Levers

The cost side is more familiar territory for AI: automating repetitive work to reduce the cost to serve and free people for higher-value tasks.

Back-office automation. Finance, operations, and administrative processes full of manual document handling and data entry, the same back-office automation that benefits the fund itself.

Customer support. AI handling routine inquiries so human agents focus on the complex and high-value ones, improving both cost and service.

Operational tasks. Sector-specific repetitive work, from scheduling to documentation to quality checks, depending on the business.

Cost levers tend to be more measurable and lower-risk than commercial ones, which makes them good candidates for the first 100-day quick win. A clear, contained automation that removes hours from a known process builds the credibility to attempt the bigger commercial bets.

7. The Data Foundation

Here is the unglamorous truth that decides most portfolio-company AI efforts: you cannot do AI value creation on bad data, and most mid-market companies have bad data.

Disconnected systems, inconsistent definitions, manual spreadsheets standing in for a database, no single source of truth for customers or products. Before AI can improve pricing, it needs clean pricing and transaction data. Before it can prioritize leads, it needs a working CRM. The data foundation is the prerequisite, and addressing it is often the highest-value early move even though it does not look like an AI project.

The practical implication for the VCP: treat the data foundation as the first AI investment, not a blocker to complain about. Fixing it unlocks every later use case and is often valuable in its own right, because better data improves decisions across the business regardless of AI. The companies that get AI value creation right are usually the ones that fixed their data first.

8. Sequencing: Quick Wins vs Platform Bets

Two kinds of AI initiative belong in the VCP, and they need different timing.

Quick wins. Contained, low-risk automations that show value in weeks: a back-office process automated, a clear productivity gain. Their job is partly the EBITDA and partly the belief; an early visible win earns the room to do the harder things.

Platform bets. Bigger initiatives that take longer and depend on the data foundation: a pricing capability, a commercial engine rebuild. Higher return, higher risk, and only sensible once the foundation and the credibility exist.

The sequencing that works: a quick win early to build momentum and trust, the data foundation in parallel, and the platform bets later on that base. The sequencing that fails: leading with an ambitious platform bet that takes a year, shows nothing for months, and loses the room. Order matters as much as choice.

9. Measuring AI Value Creation

The measurement trap is counting activity instead of value. "We deployed AI in three departments" is activity. "Pricing AI added 120 basis points of gross margin" is value. The VCP should hold AI to the second standard.

Tie every AI initiative to a value creation metric that flows to EBITDA: margin points from pricing, cost removed from a process, revenue per salesperson, hours saved valued at loaded cost. Set the baseline before you start, so the improvement is provable at exit, when a buyer will ask what actually changed.

This discipline also protects against the AI theater that is everywhere right now. A portfolio company can spend a year "doing AI" with nothing on the P&L to show for it. Measuring against EBITDA from day one keeps the effort honest and focuses it on the initiatives that genuinely move the value, which is the entire point of putting AI in the VCP. The connection of AI effort to enterprise value is explored further in our work on EBITDA and AI.

10. The Adoption Problem in Portfolio Companies

This is the factor that decides whether the AI plan delivers, and the one most VCPs underweight. A technically sound AI initiative fails if the people in the company do not use it. And portfolio company teams have every reason not to: they did not choose it, they are busy, and a new owner pushing technology is easy to wait out.

Adoption is a behavioral problem, not a technical one. It depends on involving the people who will use the tool in designing it, choosing initiatives that solve a problem the team actually feels, training on the specific tasks rather than the features, and making the new way genuinely easier than the old. A tool that is harder than the status quo will lose, every time, no matter how good the technology.

This is where the operating partner's job is hardest and most valuable. The financial case for an AI initiative is usually the easy part. Getting a portfolio company's people to change how they work is the part that determines whether the EBITDA actually shows up.

Designing AI that people in the business will actually adopt is the core of how we work, and the through-line of our portfolio deployment playbook.

11. Where to Start

Building AI into the VCP, in order.

First. Run an AI-readiness diagnostic in the first 30 days: data, value opportunities, team capability.

Second. Pick two or three use cases tied to specific EBITDA levers, fix the data gaps they depend on, and set baselines.

Third. Launch one quick win to build belief, plan the platform bets for later, and put real effort into adoption.

A Discovery Sprint runs exactly this diagnostic for a portfolio company and returns a board-ready AI plan tied to the value creation levers, with adoption built in from the start.