Most PE deals now close with an AI thesis attached, with operating partners building AI-driven margin improvement into their value-creation plans and CFOs modelling the productivity uplift before the board signs off on the investment. Within twelve months of close, most portfolio companies have an AI initiative running.

‍

What rarely happens is the question that determines whether any of it will move the exit valuation: Is this an efficiency project or an operating model change?

‍

That distinction sounds academic. In practice, it's the difference between AI investment that compounds into multiple expansions and AI investment that produces a one-time cost saving, the next buyer will already have priced in.

‍

Efficiency Doesn't Multiply

‍

The default AI use case in a PE-backed business is cost reduction, usually applied to whichever back-office process or customer-facing workflow looks most automatable on a spreadsheet. These projects are easy to scope and measure, and they tend to produce exactly the kind of value that doesn't survive a sale process.

‍

When the next buyer assesses the business, they don't pay for efficiency that's already been captured. They pay for businesses that have been structurally repositioned to operate differently. A portfolio company that has automated 30% of its support function looks like a business with a lower cost base. A portfolio company that has rebuilt how it serves customers, monetises data, or makes pricing decisions looks like a different category of asset.

Most PE-backed AI projects produce the first outcome and present it as the second. The gap between those two things is where value creation either happens or doesn't.

‍

As Kirsty Rutter, Head of Venture Capital at Lloyds Banking Group, has observed:

‍

"If you've got an organisational model that is still designed for a linear process, what you're going to get by putting AI on top of it is some efficiency. The design premise has to be what outcome do you want, not what process do you currently run."

‍

[Episode 49 – Leaders and Founders Podcast]

‍

In a PE context, that observation has a sharper edge. Efficiency gains are measurable, but they don't change what the business is. The investment thesis at exit usually requires the business to be different at exit than it was at acquisition. AI deployed against existing processes rarely delivers that.

‍

The PE firms that get this right treat AI as a deal-stage diligence question. They first define what the business could look like under a different operating model, then evaluate which AI tools support that future state. The order matters: after eighteen months of executing a plan built around today’s operating model, it’s hard to redesign the company into something fundamentally different.

‍

The strongest AI value-creation cases in PE portfolios share a common feature. The decision about how AI would change the operating model was made at the deal stage, not after close.

‍

This sounds obvious, and almost no one does it. The deal team focuses on the financial model and the commercial case. The 100-day plan is built around quick wins and integration. AI shows up in the value-creation plan as a thematic line item without a clear operating model implication. By the time the management team is engaged on what AI should do, the timeline pressure of the hold period has already started to compress the conversation.

‍

The result is that AI gets applied to the business as it already operates, rather than being used to change how it operates. Redesigning the operating model is slower, harder and requires authority that the hold-period timeline rarely allows for. Adding AI to existing processes is faster and easier to defend, so that is what gets done. The bigger value sits in the work that doesn't get scoped.

‍

The PE firms that do this well assess what the business could look like with a different operating model before they assess what AI tools could be added to the current one. A portfolio company can't redesign itself into a different shape after eighteen months of executing against a plan that assumed the original shape.

‍

What Does Foundation Readiness Mean?

‍

The other reason PE-backed AI initiatives underperform is that the technical foundations rarely support the ambition stated in the value-creation plan.

‍

This is the same problem we covered in Issue 3, but in a PE-backed business, it has a different commercial weight. The hold period is finite. The cost of discovering mid-hold that the platform can't support the AI strategy isn't just the cost of remediation; it's the cost of the value-creation thesis that no longer applies.

‍

Lee Provoost, CTO of Flagstone, made this point directly in his appearance on Leaders and Founders:

‍

"There are so many basics that companies tend to forget. If you actually get your house in order, a lot of companies could easily get a 20 to 30 percent productivity boost without even touching AI or GitHub Copilot or any of those products."

‍

[Episode 50 – Leaders and Founders Podcast]

‍

For a PE-backed business, that observation matters in two directions. The 20 to 30 percent productivity uplift Provoost describes is exactly the value PE is hoping to extract through AI. If a foundational uplift can deliver it without AI, that's the cheaper and faster route. If the platform needs AI to deliver beyond that ceiling, the foundations have to be ready, or the investment compounds the problem rather than solving it.

‍

The portfolio companies most at risk are the ones where the AI thesis assumes a clean platform that doesn't exist. Hardcoded logic, undocumented systems, and fragmented data, the same constraints that show up in a technical due diligence at exit are the constraints that prevent AI from delivering the value-creation case during the hold.

‍

What does Good Look Like in Portfolio AI?

‍

The PE-backed businesses getting compounding value from AI share a small number of characteristics.

‍

The investment thesis names a specific operating model change, not a technology adoption target. The board doesn't approve "an AI initiative." It approves a decision to run a function differently, with AI as the means to that end. That framing forces the harder conversation about whether the existing process should exist at all, which is where most of the value sits.

‍

The technical foundation is assessed before the AI scope is fixed. The business knows what its data looks like, what its architecture can support, and where the constraints sit before it commits to a roadmap. That assessment usually surfaces work that needs to happen first. The firms that get value from AI fund that foundational work as part of the AI programme, rather than treating it as a separate cost.

‍

Governance is built into the design. In regulated industries that is non-negotiable, but it matters everywhere. AI systems deployed without governance from the start create exposure that the next buyer will discount for or refuse to absorb. The firms doing this well build risk and control into the architecture rather than appending it later.

‍

Finally, technical leadership has the authority and the brief to push back on the value-creation plan when the foundations don't support it.

‍

The Next Eighteen Months

‍

The PE portfolios that have already locked their AI value-creation plans for current holdings will start to see the gap between thesis and delivery this year and next. Some will close that gap by reframing what their AI initiatives are delivering. Others will discover at exit that the buyer is pricing the work as efficiency and not paying for it.

‍

For deals being assessed now, the question is whether the AI thesis is being underwritten on a clear operating model change with a foundation that can support it, or on a thematic assumption that AI will produce margin improvement somewhere in the hold period. The first is investable. The second is increasingly visible in technical due diligence and increasingly easy to discount.

‍

The questions that determine whether AI investment will move the exit valuation are almost always easier and less expensive to answer before the deal closes than at any point afterwards.

‍

How We Help

‍

If you're scoping or executing an AI initiative, understanding whether the platform is ready should happen before delivery begins.

‍

Take our & Legacy Modernisation Assessment to identify where architectural and operational constraints may impact delivery, scalability, and future value creation

‍

Legacy Modernisation Assessment | Gathered & Found

‍