If you're a partner at a lower mid-market private equity firm, about a third of your portfolio companies probably do work that AI can already do: reading, checking, deciding, and answering. Those companies can hit their plan while competitors that build AI into the same service become faster and cheaper. Among these highly exposed companies outside software, about one in three sells an AI product when its owner talks publicly about AI. When the owner says nothing, it's fewer than one in five. In the lower mid-market, more than nine in ten of these companies are owned by a firm with no AI lead.Note 1
This is the third post in our series on AI in private equity. The firstNote 2 found that the AI hire the mega-funds make doesn't scale down to smaller firms. The secondNote 3 looked at the people who hold that job. This time we looked at what the firms own: more than 6,000 PE-backed companies, scored for how much of their work AI can already do and checked for whether they have done anything about it.
From the outside, we can't see whether a company has used AI to cut its own costs. We can see whether it has built AI into what it sells, so that is our measure of a company that has acted. Below: where exposure lands, the gap under owners who say nothing about AI, what that gap does and doesn't prove, and where to start.
A third of every portfolio does work AI can already do
We scored each portfolio company from 1 to 5 for how much of its core work AI can do today. The rubric asks two questions. First, how much of the company's time goes on reading, writing, deciding, scheduling, and answering, rather than physical work? Second, is that work repetitive enough to follow patterns? A 5 is a business whose service is itself repetitive knowledge work, such as claims processing, revenue-cycle management, or staffing. A 1 is almost all physical or asset-based work, such as mining or agriculture. Every score is tied to a specific use case in that business. In this post, a "highly exposed" company is one that scores 4 or 5.
At every size of sponsor, about a third or more of the portfolio companies are highly exposed.
Of 6,122 PE-backed portfolio companies, the share highly exposed to AI is 33% at lower mid-market sponsors (under $500M), 32% at mid-sized ($0.5–2B), 41% at large ($2–10B) and 49% at mega-funds ($10B+).
In the lower mid-market, about a third of portfolio companies are highly exposed. At the mega-funds, it's about half. Our samples for the two largest tiers lean toward firms with an AI lead,Note 4 but the point holds at every size: exposure is a property of the business, not of who owns it. A claims processor is just as exposed under a small fund as under a mega-fund.
The people who could act on that exposure are not spread the same way. By our October recount, 39% of mega-funds have an AI lead. In the lower mid-market, only 4% do.Note 3 That's why more than nine in ten highly exposed companies in the lower mid-market are owned by a firm with no AI lead.Note 1 It isn't because lower mid-market firms own unusually exposed companies. It's because almost no lower mid-market firm has an AI lead.
Where exposure lands
For 5,469 of the companies we scored, we could also check whether they market an AI product or feature. The rest of this post uses those companies.Note 5
Software is 95% highly exposed, and three in four of those sell AI. Financial services is 78% highly exposed, with 18% of those selling AI; IT and business services 60% and 39%; healthcare 34% and 20%; education 23%. Only about one in twenty industrials, energy or consumer companies is highly exposed.
- Highly exposed, sells AI
- Highly exposed, no AI product
- Less exposed
Software is the obvious case. Almost every software company is highly exposed, and three in four of those already sell an AI product. Industrial, consumer, and energy companies are at the other end. Their work is mostly physical, and almost none of them are highly exposed.
What does a score look like in practice? Pick a sector to see a company at each end of the scale and one in the middle.Note 6
Example companies at AI exposure scores of 1, 3 and 5 in five sectors. In healthcare: medical office buildings score 1, hospice care 3 and denied-claim appeals for hospitals 5. No software or financial services company in the data scores 1.
1 of 5
Mostly physical work
No software company in our data scores 1. The lowest, a 2, builds robotic telesurgery hardware.
3 of 5
Video intercoms and access control for apartment buildings
The work
Sells and installs smartphone-connected door panels and access hardware across thousands of buildings, plus the app that runs them.
Where AI fits
Support tickets and install scheduling. The value is in the hardware, the installs and the network of buildings.
5 of 5
Mostly knowledge work
Freight audit and payment for global shippers
The work
Reads carrier invoices in dozens of formats, matches every line to contracts and shipment records, and chases the exceptions.
Where AI fits
Extraction and matching are the job itself. AI can take the first pass on almost every invoice, leaving people the disputes.
1 of 5
Mostly physical work
No financial services company in our data scores 1. The lowest, the 2s, include an armoured cash-in-transit and ATM service business.
3 of 5
Wealth advisory for wealthy families
The work
Advisers manage money and plan finances for high-net-worth clients, relationship by relationship.
Where AI fits
Drafting financial plans, meeting summaries and client reports. The trust and the advice stay with the adviser.
5 of 5
Mostly knowledge work
Chargeback disputes for merchants
The work
When a shopper disputes a card payment, assembles the evidence and argues the merchant’s case with the bank.
Where AI fits
Gathering the evidence and writing the response is the service itself, and it follows a pattern AI can learn.
1 of 5
Mostly physical work
Vegetation management for electric utilities
The work
Crews clear trees and brush from power lines, and restore lines after storms.
Where AI fits
Routing crews and planning work. Useful, but the work is chainsaws and bucket trucks.
3 of 5
Elevator and escalator inspections
The work
Inspectors visit buildings to test elevators and escalators for safety, then report what needs fixing.
Where AI fits
Drafting inspection reports and ranking repairs. The inspection itself still needs someone in the machine room.
5 of 5
Mostly knowledge work
Captioning and transcription
The work
Captions, transcribes and audio-describes video for universities, companies and broadcasters.
Where AI fits
Speech-to-text and translation now do the first pass, leaving people to review and correct. That pass was most of the job.
1 of 5
Mostly physical work
Medical office buildings
The work
Buys and runs medical office buildings and leases the space to physician practices.
Where AI fits
Summarising leases and screening acquisitions. The business is property. Only two healthcare companies in our data score 1, and both are landlords.
3 of 5
Hospice and palliative care
The work
Nurses, aides and chaplains caring for patients at home, from dozens of local offices.
Where AI fits
Visit notes, recertification paperwork and billing. The care itself happens in person and stays that way.
5 of 5
Mostly knowledge work
Denied-claim appeals for hospitals
The work
Clinicians and coders read patient charts, pull out the evidence for a denied claim and write the appeal letter.
Where AI fits
Reading a chart and drafting a structured argument from it is what current models do well. A first draft of every appeal is within reach.
1 of 5
Mostly physical work
Natural gas gathering and processing
The work
Pipelines and processing plants that move gas from the wellhead to market.
Where AI fits
Predictive maintenance and spotting flow anomalies. Useful, but the business is steel in the ground.
3 of 5
Gas compression parts distributor
The work
Stocks tens of thousands of compressor parts and sells them to field operators.
Where AI fits
Quoting, cross-referencing part numbers and order entry. The warehouse, the trucks and the field service stay.
5 of 5
Mostly knowledge work
Engineering staffing for energy companies
The work
Finds engineers and technology contractors and places them in roles at energy companies.
Where AI fits
Sourcing, screening and matching candidates to roles, which is most of the service. It is the only energy company in our data to score 5.
The middle is where the risk sits. In financial services, nearly four in five companies are highly exposed, yet fewer than one in five of those sell an AI product. In IT and business services, three in five are highly exposed, and about two in five of those sell one. In healthcare, about a third are highly exposed, and one in five of those sell one. These businesses run on reading, checking, deciding, and answering, the work AI does best. Most of them haven't built that into what they sell, which leaves room for a competitor that does.
Under a silent owner, exposed companies are about half as likely to have acted
Next we took the highly exposed companies outside software and grouped them by what their owner says publicly about AI. In our first post, we sorted every sponsor into one of four stances: a named AI lead, a described AI initiative, a mention of AI in its thesis or materials, or nothing at all.Note 2 We know the stance of 508 sponsors in our data. Of those, 226 say nothing about AI.
Of highly exposed non-software companies, 34% sell an AI product when their sponsor talks about AI, against 18% when it is silent: 20% against 9% in financial services, 23% against 9% in healthcare, and 46% against 19% in IT and business services.
- Sponsor talks about AI
- Sponsor silent on AI
When a sponsor talks about AI, 34% of its highly exposed non-software companies sell an AI product. When a sponsor says nothing, it's 18%, about half as many. The same pattern shows up in every sector with enough companies to check, though each sector on its own is a small sample. In financial services and in healthcare, the share under a silent owner is less than half the share under one that talks about AI. In IT and business services, it's about one in five against nearly half.
Silent sponsors differ from the others in ways that could explain the gap on their own: they rarely focus on software, and they tend to be smaller. So we held those differences constant: sector, exposure score, company size and age, and the sponsor's size and software focus.Note 7 The gap narrows but stays wide. Under a sponsor that talks about AI, 33% of highly exposed companies sell an AI product. Under a silent sponsor, it's 20%. Put another way, a highly exposed company with a silent owner has less than half the odds of selling AI. The result holds if we drop any single sponsor, leave out the mega-funds, or count announced AI programs as well as products.
Software companies show the same pattern, but more weakly. Under a sponsor that talks about AI, 77% of software companies sell AI. Under a silent one, 69% do. Once we hold the same factors constant, the software gap could be chance.
What the gap does and doesn't show
It would be neat if the share of companies selling AI rose step by step with the sponsor's commitment, from a mention to an initiative to a named lead. It doesn't. At firms that only *mention* AI, 36% of highly exposed companies sell an AI product. At firms with a named AI lead, the share is 34%. At firms with a described initiative, it's 25%.
So the honest reading isn't that hiring an AI lead makes portfolio companies build AI products. It's that silence is a signal. A sponsor that has nothing to say about AI in public is unlikely to be pushing its most exposed companies to do anything with it either.
Cause and effect could also run the other way: a firm that already owns companies selling AI has a reason to talk about AI. Our data can't tell those two explanations apart. What it can show is where to look: highly exposed companies whose owner says nothing about AI. There are 334 of those companies in our data. Of those, 273 have no AI product.
What this means
For a lower mid-market firm, the useful question isn't what to say about AI. It's which of your companies are highly exposed and have no answer yet. A highly exposed company with no answer can hit its plan right up until a competitor that has built AI into the same service starts winning on speed and price. The same exposure is an opening for the company that moves first, because it can sell that advantage.
Most lower mid-market firms have no AI lead to build that list. Start with the companies whose work is mostly reading, checking, deciding, and answering. Rank them by where removing a bottleneck would most change growth and margin, and begin with the one at the top.
That's where Midici starts. Our portfolio assessment ranks your companies by how exposed each one is and where removing a bottleneck would move EBITDA most. Then our engineers work inside the companies that rank highest, alongside the people who do the work.
A firm doesn't need a public AI strategy to begin. It needs to know which of its companies are exposed, and to put someone on them before a competitor does.
How we did this
The companies. We mapped the current portfolios of 520 PE sponsors from our dataset, which is built from SEC Form ADV filings.Note 8 Among the largest firms, those with an AI lead are over-represented. We kept the PE-backed operating companies each sponsor still holds and scored 6,128 of them. Six are at sponsors whose size we don't know, so the firm-size figure leaves them out.
Exposure. A research agent scored each company from 1 to 5 for how much of its core work current AI can do or change, using the rubric described above: the share of time spent reading, writing, deciding, scheduling, and answering, and how repetitive that work is. When a company fell between two scores, we took the lower one. "Highly exposed" means a score of 4 or 5.
AI product. Research agents checked each company's website, LinkedIn, job postings, and press coverage, and recorded whether it markets AI as a feature of what it sells. A reviewer checked every batch of results before anything was recorded, overriding a score or a finding the evidence didn't support. We recorded only each company's strongest AI signal, and an announced AI program in the company's own operations ranks above an AI product. So a company with both an announced program and an AI product counts as not selling AI, which makes these shares floors. The agents also collected a weaker signal, AI used in a company's own operations, and we deliberately leave it out: from the outside, it's too weak to separate from noise.
Sponsor size and AI leads. From our first and second posts, recounted in October 2026. "Lower mid-market" means under $500 million in PE assets, which folds in the smallest firms that our first post counted separately. The firm-size figure uses all 6,122 scored companies whose sponsor's size we know. The rest of the post uses the 5,469 we could also check for an AI product.
Sponsor stance. From our first post: a named AI lead, a described AI initiative, a mention of AI in the firm's thesis, content, or portfolio materials, or nothing.
Analysis. Adjusted figures come from a logistic regression, a standard model for yes-or-no outcomes, on the 1,166 highly exposed non-software companies whose sponsor's stance we know. Standard errors are clustered by sponsor.Note 7 We checked it by leaving out each sponsor in turn, leaving out the mega-funds, and counting announced AI programs as well as products.
This measures public evidence. A company that builds AI quietly looks to us like a company with none, so these figures are floors. Smaller companies also publish less about themselves. These are associations, not proof of cause.
Notes
Of the 571 highly exposed companies at lower mid-market sponsors whose portfolios we mapped, 531 (93%) are at a firm with no AI lead. We leave out three large firms, Apax, Deerfield and Strategic Value Partners, that file through small US entities. Firms with an AI lead are over-represented among the portfolios we mapped, so the true share is likely higher.
Midici, “The AI hire that works for mega-funds doesn’t scale down,” September 29, 2026. https://midici.com/blog/the-ai-gap-in-private-equity-is-a-size-gap/
Midici, “Smaller PE firms rarely have a senior AI lead working on their companies,” October 12, 2026. https://midici.com/blog/who-leads-ai-at-private-equity-firms/
We mapped mega-fund portfolios mostly for firms with an AI lead, and our large-firm sample over-represents them (42 of the 43 large firms with an AI lead, against 54 of the 179 without).
Of the 6,128 companies we scored, these are the 5,469 whose websites and job postings we could also check. Sectors come from each company’s categories in our dataset; the 532 companies with none are left out of the sector figures but included in the totals.
Examples are drawn from our dataset of scored PE-backed companies and anonymised. Descriptions are paraphrased from each company’s public materials. Sectors here are assigned from the same categories, a little more narrowly than in the sector figure.
Logistic regression of “sells an AI product” on whether the sponsor is silent on AI, controlling for the company’s sector, exposure score, headcount band and age band, and for whether the sponsor focuses on software and has $2 billion or more in PE assets. Standard errors are clustered by sponsor. Silent sponsor: odds ratio 0.44 (95% CI 0.30–0.65), n = 1,166 companies at 314 sponsors. Shares quoted in the text are the model’s average predictions with every company set to each stance in turn.
U.S. Securities and Exchange Commission, Form ADV filings, Part 1 and Schedule D (private fund reporting).

