Reinventing traditional service businesses through the Generative AI era.

Explore our approach! 

Reinventing Traditional Businesses Through Generative AI

Reinventing traditional service businesses through the Generative AI era. Explore our approach!

Inflection Ventures partners with legacy service businesses - call centres, BPOs, managed services - to drive 10× value creation in less than 5 years, by shifting from labor-first to AI-first.  

What PE Firms Should Check Before They Invest

What PE Firms Should Check Before They Invest

August 12, 2026

⏱ 5 min read

by Inflection Ventures

August 12, 2026︱⏱ 5 min read︱by Inflection Ventures

Shares

The deal math has changed, and AI is being asked to close the gap. In a typical 2015 buyout, half the purchase price was borrowed at 6–7% and rising multiples did much of the work, so a deal needed only about 5% annual EBITDA growth to return 2.5x over five years. Borrowing now costs 8–9%, leverage sits closer to 30–40%, and purchase multiples are high but flat. The same return now requires 10–12% (Bain, Global Private Equity Report 2026). Bain's shorthand for the era is blunt: 12 is the new 5.

That is the pressure AI is expected to relieve, which is precisely why the diligence question about it has stopped working. In McKinsey's latest global survey, 88% of organisations report regular AI use in at least one business function, yet only 39% attribute any level of EBIT impact to it, and most of those put the figure below 5% (McKinsey, Nov 2025). Asking a target whether it uses AI now separates almost nothing. The useful questions are how much value the AI could produce, and what exposure it has already created.

The deal math has changed, and AI is being asked to close the gap. In a typical 2015 buyout, half the purchase price was borrowed at 6–7% and rising multiples did much of the work, so a deal needed only about 5% annual EBITDA growth to return 2.5x over five years. Borrowing now costs 8–9%, leverage sits closer to 30–40%, and purchase multiples are high but flat. The same return now requires 10–12% (Bain, Global Private Equity Report 2026). Bain's shorthand for the era is blunt: 12 is the new 5.

What you will find in this article:

• Why a yes on AI adoption no longer differentiates a target

• The five checks a real AI diligence process runs, in sequence

• The three failures that push AI discovery past the signing date

• How the same five checks read to an operator and to an investor

That is the pressure AI is expected to relieve, which is precisely why the diligence question about it has stopped working. In McKinsey's latest global survey, 88% of organisations report regular AI use in at least one business function, yet only 39% attribute any level of EBIT impact to it, and most of those put the figure below 5% (McKinsey, Nov 2025). Asking a target whether it uses AI now separates almost nothing. The useful questions are how much value the AI could produce, and what exposure it has already created.

What you will find in this article:

• Why a yes on AI adoption no longer differentiates a target

• The five checks a real AI diligence process runs, in sequence

• The three failures that push AI discovery past the signing date

• How the same five checks read to an operator and to an investor

Shares

The Governing Idea

The Governing Idea

Diligence does not price the AI a business has. It prices the AI a business can be made to produce, and the exposure it has already accumulated.

Bain describes the prevailing practice unsparingly: due diligence is largely defensive, an exercise in confirming what is in the CIM and building a model conservative enough for a lender and aggressive enough for the sponsor. Its prescription is full potential diligence , a multidisciplinary read on what the asset could truly be worth, with AI and digital named as one of its areas of inquiry alongside commercial, operational, tech and sustainability.

That is the right instruction. What it does not supply is the content of the AI section, and the content is where diligence usually fails. A tool count tells you what was bought; it says nothing about what work has been removed, what data that work depends on, or what happens to that data when an unapproved tool touches it.

Diligence does not price the AI a business has. It prices the AI a business can be made to produce, and the exposure it has already accumulated.

Bain describes the prevailing practice unsparingly: due diligence is largely defensive, an exercise in confirming what is in the CIM and building a model conservative enough for a lender and aggressive enough for the sponsor. Its prescription is full potential diligence , a multidisciplinary read on what the asset could truly be worth, with AI and digital named as one of its areas of inquiry alongside commercial, operational, tech and sustainability.

That is the right instruction. What it does not supply is the content of the AI section, and the content is where diligence usually fails. A tool count tells you what was bought; it says nothing about what work has been removed, what data that work depends on, or what happens to that data when an unapproved tool touches it.

The five checks below run in order. Each produces the input the next depends on, and each narrows the distance between what a target says about its AI position and what your model should assume.

The five checks below run in order. Each produces the input the next depends on, and each narrows the distance between what a target says about its AI position and what your model should assume.

The Five Checks

The Five Checks

The Three Failures That Push AI Past the Signing Date

Nearly every AI surprise that surfaces after close traces back to one of three failures during diligence.

The Three Failures That Push AI Past the Signing Date

Nearly every AI surprise that surfaces after close traces back to one of three failures during diligence.

The pattern underneath all three: AI was treated as a technology topic to review rather than a value and risk item to price. Each failure is avoidable at a cost far below what it returns, which is the part that stings in the post-mortem.

Two Lenses on the Same Five Checks

The five checks read differently depending on where you sit, but they are the same five checks.

If you run the business, they are a readiness audit. They tell you which of your own AI claims would survive a buyer walking the workflow behind them, where your data would fail at volume, and which informal tools are holding information you would not want surfaced in a data room. Every finding is measured against your own baseline: cost to serve, cycle time, capacity per head.

If you invest in or acquire businesses, they are an underwriting sequence. They tell you what the estate really contains, what the data can support, how much of the cost base is structurally removable, and how much of the AI narrative has reached earnings. Checks 01 to 03 establish the position; Checks 04 and 05 turn it into a number in the model.

Two findings should then shape how the plan is resourced. Large enterprises led on pilot volume but recorded the lowest pilot-to-scale conversion, taking nine months or more where mid-market companies moved from pilot to full implementation in roughly 90 days; and external partnerships succeeded about twice as often as internal builds (MIT NANDA). For a mid-market thesis executed with a partner rather than a hired team, that is the shape of the opportunity.

At Inflection Ventures these are not separate exercises. Diagnostics like this are built on frameworks tested through our own AI Value Accelerator engagements and inside businesses we own and operate, and delivered with VisionGroup on the execution side, because the only honest test of a diligence process is signing on the strength of it. The goal is not to catalogue AI tools. The goal is a defensible price.

The pattern underneath all three: AI was treated as a technology topic to review rather than a value and risk item to price. Each failure is avoidable at a cost far below what it returns, which is the part that stings in the post-mortem.

Two Lenses on the Same Five Checks

The five checks read differently depending on where you sit, but they are the same five checks.

If you run the business, they are a readiness audit. They tell you which of your own AI claims would survive a buyer walking the workflow behind them, where your data would fail at volume, and which informal tools are holding information you would not want surfaced in a data room. Every finding is measured against your own baseline: cost to serve, cycle time, capacity per head.

If you invest in or acquire businesses, they are an underwriting sequence. They tell you what the estate really contains, what the data can support, how much of the cost base is structurally removable, and how much of the AI narrative has reached earnings. Checks 01 to 03 establish the position; Checks 04 and 05 turn it into a number in the model.

Two findings should then shape how the plan is resourced. Large enterprises led on pilot volume but recorded the lowest pilot-to-scale conversion, taking nine months or more where mid-market companies moved from pilot to full implementation in roughly 90 days; and external partnerships succeeded about twice as often as internal builds (MIT NANDA). For a mid-market thesis executed with a partner rather than a hired team, that is the shape of the opportunity.

At Inflection Ventures these are not separate exercises. Diagnostics like this are built on frameworks tested through our own AI Value Accelerator engagements and inside businesses we own and operate, and delivered with VisionGroup on the execution side, because the only honest test of a diligence process is signing on the strength of it. The goal is not to catalogue AI tools. The goal is a defensible price.

Book a Free Strategy Session!

Book a Free Strategy Session!

AI diligence is no longer a technology review. It is the AI section of full potential diligence: upside sized and exposure priced before the number is agreed.

Adoption has already reached almost every target you will look at. What has not spread is value capture, and the distance between the two is where the next few years of returns will be decided.

AI diligence is no longer a technology review. It is the AI section of full potential diligence: upside sized and exposure priced before the number is agreed.

Adoption has already reached almost every target you will look at. What has not spread is value capture, and the distance between the two is where the next few years of returns will be decided.

The best time to find your AI upside is before you sign, not after. Bring us a live deal or a portfolio company and we will run the five checks in the order an investment committee would, then name the one that changes your number. Explore the two tracks at www.inflectionventures.co or connect with the Inflection Ventures team!

The best time to find your AI upside is before you sign, not after. Bring us a live deal or a portfolio company and we will run the five checks in the order an investment committee would, then name the one that changes your number. Explore the two tracks at www.inflectionventures.co or connect with the Inflection Ventures team!

← Back to Blog

Copyright © 2026 Inflection Ventures. All rights reserved.

Copyright © 2026 Inflection Ventures. All rights reserved.