Down from about 31 minutes to
1 min
for the first reply on its support desk, since it put an AI agent on it
29% of companies owned by private equity show AI in what they sell or how they run. Outside software, it's 17%.
We read the websites of nearly 9,000 companies in the portfolios of US private equity firms, for what each one says about AI.
- More than 700
- US private equity firms whose portfolio pages link to their companies
- More than 14,000
- companies on those pages with a website
- Nearly 9,000
- company websites we read
Key findings
- 29% of companies owned by US private equity firms show AI at work on their own websites, in what they sell or in how they run.
- Most of that is software. 79% of software and IT companies show AI at work, against 17% of every other company.
- After software, business services show it most, at 43%. Consumer products show it least, at 3%, and retail and restaurants are at 4%.
- 37% of the companies' sites say anything about AI.
- 3 to 4% say they use AI in how they run their own work. Fewer than 1 in 1,000 say what it changed, with a number or a before and after.
- Where the owner says AI is part of how it improves its companies, 34% of them show AI at work. For companies in the same industries with other owners, it's 28%.
- Companies of firms with more than $10 billion in private funds show AI at work most often, at 34%. At firms under $2 billion it's 26 to 27%, comparing companies in the same industries.
Get the next study by email
Our new studies by email: what we found and the data behind it.
Unsubscribe from any email. How we use your address
Outside software, 17% of companies show AI at work.
79% of the software and IT companies owned by private equity show AI in what they sell or how they run. Business services come next, at 43%. In consumer products it's 3%.
- Software and IT
- 79%
- Every company outside software
- 17%
- Business services
- 43%
- Transportation and logistics
- 33%
- Education
- 32%
- Aerospace, defense and government
- 30%
- Financial services
- 30%
- Media and telecom
- 24%
- Life sciences and medical products
- 19%
- Healthcare services
- 17%
- Energy and environmental
- 12%
- Automotive
- 7%
- Industrial products
- 6%
- Construction and home services
- 5%
- Retail, restaurants and leisure
- 4%
- Consumer products
- 3%
A company shows AI at work when its site says that what it sells uses AI, or that it uses AI in its own work. The dark bar is every company outside software together.
fulkersonadvisors.com
Where the owner puts AI in its playbook, more of its companies show it.
34% of those companies show AI at work, against 28% of companies in the same industries with other owners. What each firm says comes from our reading of US private equity firms' own websites.
- Say anything about AI
- Owner puts AI in its playbook: 42%Other owners: 35%
- Show AI at work
- Owner puts AI in its playbook: 34%Other owners: 28%
Both groups are weighted to the same mix of industries, so an owner of more software companies doesn't come out ahead for that alone. Before that weighting, 44% and 24% show AI at work.
fulkersonadvisors.com
Companies of the biggest firms show the most AI at work.
At firms with more than $10 billion in private funds, 34% of companies show AI at work. At firms under $2 billion it's 26 to 27%. Each size is weighted to the same mix of industries.
- Under $100M: 27%
- $100M to $500M: 26%
- $500M to $2B: 27%
- $2B to $10B: 31%
- Over $10B: 34%
fulkersonadvisors.com
Seven companies say what AI changed in their own work.
3 to 4% of the companies say they use AI in how they run their own work. Fewer than 1 in 1,000 put a number or a before and after on what changed. These do, in their own words.
20
full-time employees' worth of patient prescreening, done by its AI recruiting tool
$1M
in savings from its own 90-day AI pilot, before it sold the program to clients
Ducker Carlisle
Business services, backed by Boathouse Capital
“As a PE-backed firm, we ran our own 90-day AI citizen developer pilot before selling it to clients. We deployed a decentralized model that enabled teams across finance, operations, and business development to build their own AI solutions. Result: $1M in validated operational savings and a scalable playbook we now offer to portfolio companies.”
3.5x
as many merged pull requests a week on one of its own teams, eight months into using its AI tool
40%
of submissions for its malpractice insurance now get an automated underwriting decision
About
10%
better fuel efficiency on its own freight, in a pilot of AI-connected trucks
EASE Logistics
Transportation and logistics, backed by The O.H.I.O. Fund
“During this pilot, two semi‑trucks traveled approximately 44,000 miles – over 11,000 in platooning mode – achieving about a 10% boost in fuel efficiency and offering valuable insights into technology performance in rural terrain.”
Without a number
PS Logistics
Transportation and logistics, backed by Gamut Capital Management
“P&S Transportation, a leading flatbed carrier and subsidiary of PS Logistics, today announced the successful deployment of VIC (Virtual Information Coordinator), an industry-leading AI logistics coordinator capable of handling complex conversations in real-time. Developed in partnership with Hyperscale Systems, VIC has eliminated hold times at scale with the same informational awareness as other dispatch channels.”
Putting AI to work in a company you own?
We build AI systems that companies use every day, and the company owns the software. If you run a company owned by private equity, or invest in one, talk it through with us on a thirty-minute call.
The data.
We read the sites on October 29, 2026.
What the owners say about AI is in our reading of US private equity firms' own websites.
How we did it
We started from the 1,115 US private equity firms in our reading of private equity firms' websites and read their sites again on October 29, 2026. Jev, a model from TypeSafe that answers questions with a probability, picked each firm's portfolio page from the links on its homepage, or from its sitemap where the homepage had few links. 1,052 firms had one. We read pages as plain HTML, and in a browser where that came back nearly empty, so lists built by script were read too.
Each portfolio page became a list of cards in page order: links with every label that points to them, logos, headings and names written in the text. Clairic, a Cloudflare model that copies text out of a page, copied the company names, and a name was kept only where it appears on the page exactly. Jev judged each card, whether it is one of the firm's companies, whether the page shows it as current or past, and whether its link goes to the company's own website. Where a card led to a page on the firm's own site, Jev read that page for the company's website. 932 firms list companies and 724 link at least one to its website. Merged by web address across firms, that's 14,014 companies, and 11,318 of them aren't marked as past.
For each of those, we read the homepage and up to three more pages Jev picked from its links, on what the company sells, its technology and its news. We obeyed robots.txt, and a site that refused us stayed unread. 9,928 sites could be read. Clairic copied every passage about AI from those pages, with no list of keywords. Jev then answered every question about a company in one request. Is the site the listed company's own working site, and in which industry? Does it say anything about AI? Does what the company sells use AI, does the company use AI in its own work, does it name a result of its own AI use, and does it give customers a number for its AI? A yes counts at 50% or more. 8,896 sites were the listed company's own working site, and every share on this page is of those.
A company shows AI at work when its site says that what it sells uses AI or is AI, or that the company uses AI in how it runs its own work. AI that only comes up as an industry trend doesn't count, and neither does a partner's AI with no word that the company uses it, a privacy notice, an accessibility widget or a job post.
Firm size is the gross assets of the firm's private funds as reported to the SEC, and whether a firm puts AI in its playbook is the firms study's answer for that firm. A company counts with each firm that lists it and doesn't mark it as past. Comparisons within the same industries weight each group's share in each industry by the mix of industries across the companies read, counting an industry for a group when the group has at least 10 companies in it.
To check the reading, a second model, OpenAI's GPT-6-Astra, read 350 companies drawn from every kind of answer Jev gave, without seeing Jev's answers, and 100 links from portfolio pages. Weighted back to the companies read, the two agree on whether a company shows AI at work for 98% of companies and on whether its site mentions AI for 98%. The second model puts the share showing AI at work at 28%, against Jev's 29%. It finds AI in a company's own work less often, 3% against Jev's 4%, so this page gives that share as a range. On the portfolio-page links, the two agree on 98 of 100. We read each of Jev's 23 results by hand and kept the 7 where a company says what changed in its own work, with a number or a before and after.
This measures what companies say on their own websites. A company can use AI and not say so, and a site can say more than a company does. Firms that list their companies without linking to their websites are left out, and a page that doesn't mark exits can include companies the firm has since sold. People are never named.
The CSV has one row per company we read, with its website, its industry, the firms that list it and the size of their private funds, and Jev's answers. It's free to use under CC BY 4.0 with attribution to Fulkerson Advisors.
Cite this study
Fulkerson Advisors (2026). AI at companies owned by private equity: 8,896 company websites, October 2026. https://www.fulkersonadvisors.com/research/ai-in-private-equity-portfolio-companies
The charts and the CSV are free to use with a link to this page.
The data is also on Zenodo, DOI 10.5281/zenodo.23806110.
Questions
How many companies owned by private equity use AI?
On their own websites, 29% of the companies in US private equity portfolios show AI in what they sell or how they run, and 37% say anything about AI. A site shows what a company says, so a company can use AI without it showing up here.
Which private equity-owned companies use AI the most?
Software and IT companies, at 79%. After them come business services at 43% and transportation and logistics at 33%. Consumer products are lowest, at 3%.
Do private equity firms that talk about AI own companies that use it?
More often, yes. Where the firm says AI is part of how it improves its companies, 34% of its companies show AI at work, against 28% of companies in the same industries with other owners.
Do portfolio companies say what AI changed?
Very few. 3 to 4% say they use AI in their own work, and fewer than 1 in 1,000 say what it changed with a number or a before and after. Software Toolbox and Velocity Clinical Research are two of them.
Do bigger private equity firms' companies use more AI?
They show more of it. At firms with more than $10 billion in private funds, 34% of companies show AI at work, against 26 to 27% at firms under $2 billion, comparing companies in the same industries.
How is this different from your study of private equity firms?
That study read the firms' own websites for what the owners say about AI. This one reads the websites of the companies they own, and uses that study's answer for whether each owner puts AI in its playbook.