The PE Firm's Guide to AI That Actually Works
Mar 16, 2026
|
Key Takeaways ▶ 85% of PE firms pushing AI adoption, but most see negligible EBITDA impact ▶ Three mistakes kill ROI: pilot purgatory, wrong assets, ignoring governance ▶ Highest AI ROI is in “boring” industrial assets, not tech portfolio companies ▶ Production systems in weeks, not pilot programmes that run for quarters |
|
85% of PE firms pushing AI adoption across portfolios |
70% increasing AI investment 25%+ in 18 months |
EUR 45m at-risk revenue identified in 6 weeks at a European industrial distributor |
There’s a paradox in PE’s relationship with AI. Investment is accelerating - 85% of firms are pushing portfolio companies toward adoption (CLA Connect), and 70% plan to increase AI investment by 25%+ in the next 18 months (Bain). Yet the EBITDA impact from most AI initiatives is negligible.
Bain’s 2026 GP Outlook makes this concrete: 39% of GPs don’t expect material AI financial impact on portfolio companies this year. Not because AI doesn’t work. Because most firms are deploying it wrong.
The three mistakes that kill AI ROI
|
Mistake 1: Pilot Purgatory Most PE AI initiatives start with a “landscape scan” that produces a dozen use cases. Three get approved as pilots. Six months later, one pilot is working - sort of - but it’s solving a problem the board doesn’t care about. The other two are quietly shelved. |
The fix: start with the EBITDA target, not the technology. What does the operating partner need to move? Revenue protection? Margin expansion? Operational efficiency? Define the outcome first. Then deploy AI against it. Working solution in weeks, not a pilot programme that runs for quarters.
|
Mistake 2: Deploying AI in the Wrong Assets The instinct is to apply AI to the “best” portfolio companies - the ones with clean data, modern systems, and tech-savvy management teams. These tend to be software or tech-adjacent businesses. |
But the highest AI ROI isn’t in tech portfolio companies. It’s in “boring” industrial assets - manufacturing, distribution, logistics, facilities management. These businesses have complex operations, under-managed data, legacy systems, and massive efficiency headroom. The delta between their current operations and AI-optimised operations is enormous.
OpenGate Capital understood this. Their OGx platform applies “big tech” operational tools to “small industrial” turnarounds - taking the methods that tech companies have used for years and deploying them in businesses that have never had access to that capability.
American Industrial Partners takes a similar approach. Their Industry 4.0 programme deploys operational AI across industrial assets - not because they’re the most exciting companies, but because they have the most headroom.
|
Mistake 3: Ignoring Responsible AI PE firms face increasing scrutiny on how portfolio companies use AI. ESG-sensitive LPs are asking about data governance, bias, and compliance. Exit buyers - particularly strategic acquirers and publicly-listed companies - are applying responsible AI criteria to acquisitions. |
Deploying AI without governance frameworks creates risk that compounds through the hold. The time to address this is at deployment, not at exit preparation.
Where AI actually moves EBITDA
The use cases that generate measurable EBITDA impact share three characteristics: they target high-impact workflows, they produce measurable outcomes within weeks, and they compound over time.
|
Revenue Intelligence Customer churn prediction, revenue concentration analysis, cross-sell and upsell identification. In industrial businesses, customer behaviour patterns are deeply predictable - but only if you’re looking at them. Most industrial sales teams operate on intuition and relationship. AI-driven revenue intelligence gives them visibility they’ve never had. |
|
Case in Point At a European industrial distributor, we deployed Order Book Intelligence in six weeks. Not as a pilot, but as a production system. It identified EUR 45m in at-risk revenue by analysing customer ordering patterns against historical baselines. The sales team now uses it daily. |
|
Operational Efficiency Quote automation, process optimisation, demand forecasting. These aren’t glamorous applications. They’re high-frequency workflows where small improvements compound into significant EBITDA impact. A 70-80% reduction in quote processing time doesn’t just save time - it accelerates the sales cycle and improves win rates. |
|
Knowledge Preservation Industrial businesses run on tribal knowledge. The people who know why a machine runs this way, why a customer orders on this schedule, why a supplier needs to be managed differently - that knowledge lives in their heads. When they leave - retirement, restructuring, poaching - it walks out the door. AI-driven knowledge capture systems turn that tribal expertise into searchable, accessible intelligence. For a PE firm holding an asset for six years through multiple management changes, this isn’t a nice-to-have. It’s risk mitigation. |
|
Quote automation At a global manufacturer, hundreds of RFQs arrive daily in multiple formats and languages. Manual process: 3-5 days per quote. Customers move on before they get a response. AI reads the email and attachments, identifies products, looks them up in the ERP, builds the quote, and sends it back. 80% automation rate. Quote turnaround from days to minutes. Speed to quote is speed to revenue - the sales team focuses on relationships and exceptions, not copy-pasting. |
The discipline that separates success from failure
The PE firms getting real EBITDA impact from AI share four characteristics.
Outcome-led, not technology-led. Start with the EBITDA target, not the technology. Whether the initiative is driven by the operating partner, the CTO, or the CFO matters less than whether the outcome is defined in business terms. Revenue protection? Margin expansion? Operational efficiency? Define the outcome first, then deploy AI against it.
Weeks, not months. If a solution takes six months to deploy, the business case is already stale. The best AI deployments prove value within weeks and scale from there. Speed isn’t just about efficiency - it’s about maintaining momentum and credibility with the management team.
Platform, not point solution. Individual AI tools solve individual problems. Platforms solve problems across the business and across the portfolio. Revenue intelligence, operational efficiency, and knowledge capture all drawing from the same data foundation. The compounding effect of a platform approach is what creates sustainable operational alpha.
Production, not pilot. Pilots are experiments. The teams on the ground need production systems they use daily. The gap between “interesting pilot” and “tool the sales team relies on” is where most AI initiatives die. The firms generating returns have closed that gap.
The firms that treat AI as an operational lever - not a technology initiative - are the ones generating returns. The rest are spending money on experiments that never scale.
The question isn’t whether your portfolio companies should use AI. That decision has already been made - 85% of firms are pushing adoption. The question is whether that AI is deployed against the right targets, in the right assets, with the right governance. That’s the discipline that separates measurable EBITDA impact from another round of pilots that go nowhere.
Sources & References
| CLA Connect: PE Firms and AI Adoption - 85% of PE firms pushing AI adoption across portfolios |
| Bain & Company: PE AI Investment Research - 70% increasing AI investment 25%+ in next 18 months |
| Bain & Company: Private Equity’s Reality Check - The GP Outlook for 2026 - 39% of GPs expect no material AI impact in 2026 |
| OpenGate Capital: OGx Platform - “Big tech” operational tools for “small industrial” turnarounds |
| American Industrial Partners: Industry 4.0 Programme - Operational AI across industrial portfolio assets |
Frequently Asked Questions
|
What are the biggest mistakes PE firms make with AI? Three mistakes kill AI ROI in PE: (1) Pilot purgatory - starting with landscape scans that produce dozens of use cases, leading to pilots that run for quarters without measurable EBITDA impact. (2) Deploying AI in the wrong assets - applying AI to tech-adjacent portfolio companies with clean data instead of industrial assets with massive efficiency headroom. (3) Ignoring responsible AI - deploying without governance frameworks, creating compounding risk through the hold period that surfaces at exit. |
|
What is pilot purgatory in PE AI? Pilot purgatory is the common failure mode where PE AI initiatives begin with a landscape scan producing a dozen use cases, three get approved as pilots, and six months later one is working - sort of - but solving a problem the board doesn’t care about while the other two are quietly shelved. The fix is to start with the EBITDA target, not the technology. Define the outcome first, then deploy AI against it, delivering a working solution in weeks rather than running a pilot programme for quarters. |
|
Which portfolio companies have the highest AI ROI? The highest AI ROI isn’t in tech portfolio companies - it’s in “boring” industrial assets like manufacturing, distribution, logistics, and facilities management. These businesses have complex operations, under-managed data, legacy systems, and massive efficiency headroom. The delta between their current operations and AI-optimised operations is enormous. OpenGate Capital’s OGx platform applies “big tech” operational tools to “small industrial” turnarounds for exactly this reason. |
|
Where does AI actually move EBITDA? AI moves EBITDA in four areas: (1) Revenue intelligence - churn prediction, concentration analysis, cross-sell identification. At a European industrial distributor, Order Book Intelligence identified EUR 45m in at-risk revenue in six weeks. (2) Operational efficiency - process optimisation, demand forecasting, accelerating decision cycles. (3) Knowledge capture - turning tribal knowledge into searchable intelligence before it walks out the door. (4) Quote automation - at a global manufacturer, AI cut quote turnaround from 3-5 days to minutes with 80% automation. |
|
What is revenue intelligence in PE? Revenue intelligence is AI-driven analysis of customer behaviour patterns - ordering frequency, purchasing volumes, and behavioural signals - to identify churn risk, revenue concentration, and cross-sell opportunities. In industrial businesses, customer behaviour patterns are deeply predictable but only if you’re analysing them systematically. Most industrial sales teams operate on intuition and relationship; AI-driven revenue intelligence gives them visibility they’ve never had, identifying at-risk revenue before it’s lost. |
|
What discipline separates AI success from failure in PE? Four disciplines separate success from failure: (1) Outcome-led, not technology-led - start with the EBITDA target, whether driven by operating partner, CTO, or CFO. (2) Weeks, not months - if a solution takes six months to deploy, the business case is already stale. (3) Platform, not point solution - platforms solve problems across the business and portfolio, creating compounding operational alpha. (4) Production, not pilot - teams need production systems they use daily, not experiments that never scale. |
|
Why should AI deployment be outcome-led? AI deployment should start with the business outcome, not the technology. Whether the initiative is driven by the operating partner, CTO, or CFO matters less than whether the outcome is defined in business terms - revenue protection, margin expansion, operational efficiency. The firms generating real EBITDA impact define the target first, then deploy AI against it. The technology serves the business objective, not the other way around.
|
This is Part 5 of an 8-part series on the structural shift from financial engineering to operational value creation in private equity. New articles publish weekly through April 2026.
Keep informed with the newsletter for PE operating partners and the portfolio companies they back.
Get operational insights and trends, AI frameworks, resources and real deployment stories.
