The Disruption Column: Adding AI Exposure to the Investment Committee Memo

ai disruption ai in pe ic memo risk underwriting value creation Oct 29, 2026

Investment committee memos have a stable anatomy. Thesis, market, competition, financials, management, value creation plan, risks, exit. The risk section has grown over the years by accretion. Cyclicality got a paragraph after 2009. Cyber got one after the first portfolio breach. ESG got a page when the LPs started asking.

AI exposure has not yet been added, and it is being handled in the worst possible way: as an enthusiasm rather than a risk. Most memos in 2026 mention AI as an upside lever in the value creation plan. Very few ask whether the target's revenue is standing in front of it.

The short answer

Add one column to the revenue build: for each material revenue line, what is the customer actually paying for, and can a machine do that thing. Three exposure grades follow. Displaceable revenue, where the work is the deliverable and the deliverable is now cheap. Compressible revenue, where the work survives but the price does not. Insulated revenue, where liability, relationship, physical presence, or regulation sits between the customer and the machine. The output is not a yes or no on the deal. It is a discount rate applied to a specific slice of revenue, and a hold period that assumes the slice shrinks.

Why the question is usually skipped

Three reasons, and none of them are stupid.

The first is that it feels speculative. Underwriting committees are trained to resist arguments that depend on predicting technology, and rightly so. The industry has a long record of expensive disruption theses that did not arrive.

The second is that it is uncomfortable for the deal team. A serious AI exposure analysis on a services business frequently produces an answer the sponsor does not want. It is easier to place AI in the upside column, where it improves the model, than in the risk column, where it reduces the price.

The third is that there is no standard format. Cyclicality has a stress case. Cyber has a questionnaire. AI exposure has neither, so it is discussed in the room and does not survive into the document.

The fix is not a prediction. It is a structured description of what the customer is paying for, which is a present-tense question with a knowable answer.

The three grades of exposure

Take the revenue build. For each line representing more than ten percent of revenue, ask what the customer is actually buying. Not what the company sells. What the customer would say they are paying for if asked plainly.

Exhibit 1 · Three grades of AI exposure

Grade	What the customer is buying	Underwriting treatment
Displaceable	The production of an artefact. A document, a translation, a first-pass analysis, a design, a summary. The work is the deliverable.	Assume the line declines over the hold. Do not underwrite growth on it.
Compressible	The work still happens, but a competitor doing it with machine leverage can charge less and stay profitable.	Hold volume, model price erosion. Margin is the exposed variable, not revenue.
Insulated	Liability transfer, a licensed signature, physical presence, a regulated relationship, or trust in a named person.	Underwrite normally. This is the durable revenue.

The line between compressible and insulated is where most of the real analysis sits, and it usually turns on one question: who carries the consequence if the output is wrong?

If the customer carries it, the work is compressible, because the customer will eventually do it themselves more cheaply. If the vendor carries it, contractually or professionally, the work is insulated, because the customer is buying a transfer of risk and machines do not absorb risk. That distinction survives almost every technology cycle, and it is a more reliable guide than any assessment of what the technology can currently do.

Working the exhibit on a real shape

Take a mid-market professional services business, ninety million of revenue, four lines.

Exhibit 2 · Exposure applied to a revenue build

Compliance filings and sign-off Licensed signature, liability carried		$40m · Insulated
Ongoing advisory retainers Named partner relationship		$25m · Compressible
Analysis and report production Artefact is the deliverable		$16m · Displaceable
Data and document processing Volume work, no judgment		$9m · Displaceable

Illustrative structure rather than a real company. Twenty-eight percent of revenue is displaceable and another twenty-eight percent is margin-exposed. The blended growth rate in the base case cannot be applied to all four lines.

The memo consequence is specific. A single blended growth assumption across ninety million of revenue is now indefensible, because two of the four lines are moving in a different direction from the other two. The exposure column does not kill the deal. It changes the shape of the model, and it changes what the hundred-day plan has to prioritise.

The second-order question: does the displaceable line fund the insulated one?

This is the part that gets missed, and it is where the real damage sits.

In many professional services and information businesses, the low-judgment work is not just revenue. It is the training ground. Junior staff learn the domain by doing the volume work, and the senior people who carry the insulated revenue in ten years are the ones who did the volume work today.

Automate the displaceable line and the immediate margin improves. The pipeline of people who become the insulated revenue quietly stops filling. That cost lands in year six, comfortably beyond the hold period, which is exactly why it does not appear in anyone's model.

If you are buying the business, this is a genuine argument in your favour on price and against you on the exit story, because the next buyer will be asking the same question with four fewer years of runway. The parallel inside a portfolio company, where the roles that look automatable turn out to be load-bearing, is worked through in the operating maturity index under talent maturity.

The four lines to add to the memo

The whole thing fits in half a page, which is the only version anyone will actually adopt.

Exposure split. Revenue by grade, in dollars and percent. One table.

Consequence test. For the compressible and insulated lines, one sentence each on who carries the risk if the output is wrong.

Competitor position. Is any competitor already pricing with machine leverage in the displaceable and compressible lines. If yes, the erosion has started and is measurable. If no, you are estimating timing, and should say so.

Pipeline dependency. Does the displaceable work train the people who deliver the insulated work. If yes, name the mitigation and cost it.

What this is not

It is not a prediction about artificial general intelligence, and any version of this analysis that depends on one should be rejected. Everything above is answerable from the present: what the customer buys, who carries the liability, what competitors currently charge, how people currently learn the job.

It is also not a reason to avoid exposed categories. Businesses with a displaceable slice are frequently mispriced downward by the market's general anxiety, and a sponsor who has done this analysis precisely knows which part of the revenue is actually at risk and which part is not. That is an edge, and it runs in the direction of buying rather than avoiding.

The failure mode is not owning an exposed business. It is owning one while modelling it as though every line behaves the same way, discovering in year two which lines did not, and explaining to the IC that the market moved. The market did not move. The column was missing.

CVCA Level 2

Revenue that behaves differently, modelled differently

A blended growth rate across four lines moving in different directions is the failure this article describes. CVCA Level 2 is nine quantitative modules on Northbridge Instruments, carried from Year 0 to exit: building a value bridge that reconciles, defending every number in it under questioning, and translating a return target into work. Includes three months of PE Simulator access at Professional level. $299.

Enrol in CVCA Level 2 Or find which programme fits you

VCI Institute in collaboration with Mohamad Chahine
Published 29 October 2026

Related reading from the VCI Institute

Underwriting the AI-Native Target
The same analysis run in reverse, on a company whose moat is a model.

The Pre-Close Operating Veto
Who should be forcing this column into the memo, and what authority they need.

The Six Demand-Side Drivers
How an exposed revenue line reads to the buyer at exit rather than to you at entry.

About the VCI Institute

The VCI Institute is a nonprofit dedicated to building practical capability and shared standards for value creation in private equity. The Institute publishes operator-grade frameworks and runs certification programmes for operating partners, portfolio company executives, and value creation analysts. You can see what each programme actually covers before deciding. Analysis published here draws on the Institute's certification curricula and on structured review of mid-market transaction patterns rather than on any single proprietary dataset. Where a figure is directional rather than measured, it is described as such.

Further material is available in the Institute's Insights library and its free resource library of templates, checklists, and case snapshots.

Affiliated ventures

The Institute is affiliated with two commercial ventures founded by the same team. They are named here rather than inside the analysis above.

Valuethropy builds OPOs, the Operating Partner Operating System: a self-serve system that carries the full value creation role across a mandate, with the human in command at every step.

AIVROS is an AI value creation and readiness platform. It assesses how ready a company, fund, or portfolio actually is for AI, and maps the optimal sequence for adopting AI in service of value creation.

© 2026 VCI Institute. All rights reserved. The frameworks, terminology, and analysis presented in this article are the intellectual property of the VCI Institute. Reproduction or derivative use without written permission is prohibited. Citation with proper attribution is welcomed.