Skip to content
Find my revenue leak

INSIGHT / AI IN PRIVATE EQUITY

AI in Private Equity: The 100-Day Plan for Portfolio Companies

How funds use AI in sourcing, diligence and monitoring, and the 100-day plan that turns AI into portfolio EBITDA: revenue first, then cost, with the exit math and the buyer's checklist.

By Timur GrigorchukPublished November 20, 202611 min read

The short answer

AI in private equity now works in two places: inside the fund, where it speeds sourcing and diligence, and inside the portfolio, where it must add EBITDA. Most of the value is still unclaimed.

In FTI's 2026 survey of 200 fund and operating leaders, 36% of portfolio companies use AI across use cases and 7% run it at enterprise scale. Start with revenue, then cost, in 100 days.

In this article 8 sections
Only 7% of portfolio companies run AI at enterprise scale, while 95% of funds say it meets the business case; 6 of 7 famous AI-in-private-equity numbers show no sample or named study
AI in the portfolio: 95% of funds report AI meeting the business case, 7% of portfolio companies run it at enterprise scale (FTI Consulting, 2026; Bain and StepStone, 2026).
Embed this infographic

Free to reuse under CC BY 4.0 with a link back. Copy the code; the credit line is the licence condition.

What is AI in private equity, and how far has it spread?

AI in private equity means two jobs. The first is the fund's own work: finding targets, reading data rooms, tracking portfolio numbers. The second is the portfolio's work: more revenue and lower cost inside the companies the fund owns. The second job pays the returns.

Adoption is wide and thin. EY reports that only one in ten of its PE clients had not used data analytics and AI for diligence or target search by 2024 (EY, 2025). FTI's 2026 survey of 200 fund and operating leaders puts 36% of portfolio companies on AI across use cases. Only 7% run it at enterprise scale.

Results sound good and cluster at the top. 95% of funds in the same FTI survey say their AI work meets or beats its business case. Bain and StepStone asked 103 GPs in December 2025 and January 2026. 39% do not expect AI to have a material financial impact on their portfolio companies in 2026.

Both can be true. Teams that count a result report a good one. Many companies have not counted yet. The gap between 36% and 7% is the opening for an operating partner.

Recent first. Surveys differ in who they ask. Fund leaders report on their own programs, which tilts results upward.
StatisticNumberSource
Portfolio companies using AI across use cases#36%FTI Consulting, 2026
Portfolio companies running AI at enterprise scale#7%FTI Consulting, 2026
Funds saying AI initiatives meet or exceed the business case#95%FTI Consulting, 2026
Respondents who name talent as the main barrier to scaling#35%FTI Consulting, 2026
GPs expecting no material AI financial impact on portfolio companies in 2026#39%Bain and StepStone, 2026
PE clients that had not used data and AI in diligence or target search, 2024#1 in 10EY, 2025
Fund groups that began or increased AI use in the past year#77%Deloitte 24th Fair Valuation Pricing Survey, October 2026, as reported by Evident
Investors expecting to expand portfolio digital budgets over three years#More than 90%BCG, 100 senior PE investors, 2026

How do funds use AI in sourcing, diligence and monitoring?

Inside the fund, AI has settled into the front of the deal cycle. Bain and StepStone's GPs most often name due diligence and deal sourcing as the place generative AI pays back first. These are reading and matching jobs, and models do them fast.

Diligence is the clearest case. In BCG's survey of 100 senior PE investors, 73% run digital diligence on most deals. The next steps drop off. 29% bring digital value creation into the pre-deal phase and 22% say digital readiness can change a go or no-go call. Only 11% tie digital progress to the exit story.

Monitoring is where the savings are plainest. EY describes a North American mid-cap fund that cut a reporting cycle from four person-days to under one hour with AI dashboards (EY, 2025). KPMG's 2026 interview in Private Equity International says the same shift: from pilots to EBITDA discipline, with clean, governed data a prerequisite for scaling AI.

What does this mean for a portfolio CEO? Your sponsor now reads your numbers faster, and compares them across the fund. A business that can show source, speed and conversion by week will look better against peers than one that reports a month late.

Funds check digital in diligence, then 11% carry it to the exit story
  1. Run digital diligence on most deals73%
  2. Plan digital value creation before the deal29%
  3. Let digital readiness change a go or no-go22%
  4. Link digital progress to the exit story11%

BCG, Private Equity's Future Is Digital First and AI Powered, January 2026, survey of 100 senior PE investors.

Where does AI add EBITDA inside a portfolio company?

Bain's 2026 report gives the target. A typical deal needed 5% yearly EBITDA growth for a 2.5x return over five years in the 2010s. Today it needs about 10% to 12%, with borrowing at 8% to 9% and record entry prices leaving less room for leverage and multiple expansion. Our private equity portfolio page covers that math. AI has to help close that gap.

Revenue comes first because it compounds. A lead answered in five minutes, a quote followed up on day three and a call that books the visit all add revenue at a high flow-through. The conversion data is old and consistent. Leads reached in the first five minutes convert 8 times more often (InsideSales, 5.7 million leads, 2021). Only 0.1% of leads get that reply. In 60 million tracked calls, 37% of phone leads converted during the call (Invoca, 2025).

Cost comes second because it is real and finite. Roland Berger's portfolio examples show the size: a 7% saving on indirect spend in under 10 weeks, forecast errors down 25%, unplanned downtime down 18% (Roland Berger, 2025). These are client examples, not averages, and Roland Berger gives no scope for the forecast figure. They show what a focused sprint can do on one cost line.

Funds already lean toward the cheaper-to-prove side. Bain and StepStone report that portfolio outcomes skew toward cost savings and efficiency more than revenue growth. FTI finds 41% of respondents call revenue acceleration their top priority. The priority is revenue. The results so far are cost. That is the gap this plan closes.

  • REVENUE SIDE FIRST

    Four places AI adds revenue

    Lead response
    Every form, call and chat gets a first reply in minutes, day and night. Evidence: 8x conversion inside five minutes (InsideSales, 2021).
    Follow-up
    Every open quote gets a next step and a date. A person owns each one. AI drafts, a person sends.
    Call analysis
    Every recorded call is read for objections, price talk and booked next steps. Coaching follows the calls that lost.
    Pricing
    Discount and price decisions read from past wins. Roland Berger cites one dynamic pricing model worth $5M of margin in a quarter (2025, single case).
  • COST SIDE SECOND

    Four places AI cuts cost

    Procurement
    Indirect spend read line by line. Roland Berger cites 7% saved in under 10 weeks (2025, single case).
    Customer support
    Common questions answered in the first reply. A person takes the rest.
    Finance close
    Reporting cycles shrink. EY cites four person-days down to under one hour (2025, single fund).
    Operations
    Forecasts and maintenance flags. Roland Berger cites forecast error down 25% and unplanned downtime down 18% (2025).
Examples are cited as single cases, not benchmarks. Your base rates decide your result.

What does +20% conversion earn in EBITDA and exit value?

A conversion rate is a revenue number, and revenue becomes enterprise value through EBITDA. Here is an example, not a benchmark. A portfolio company has $60M of revenue, a 15% EBITDA margin and $9.0M of EBITDA. $20M of its revenue comes from new customers who first arrived as inbound leads.

Lift the lead-to-customer rate from 2.5% to 3.0%. That is 20% more customers from the same leads, with no extra ad spend. Revenue rises by $4.0M. At a 35% flow-through, which is the share of new revenue that reaches EBITDA, that adds $1.4M. Take off $0.3M a year to run the AI and the people who own it. Net EBITDA rises $1.1M.

At a 10x exit multiple, $1.1M of EBITDA is $11.0M of enterprise value. The fund spent months, not years, to earn it. The same lever also covers Bain's gap for one year: EBITDA up 12.2% from this single change.

Change the inputs for your own company. A rate that starts higher or a margin that starts lower moves the answer. The point is the shape: a small lift on a large lead flow lands as a visible slice of EBITDA, and the multiple turns it into value.

Example. Net EBITDA gain of $1.1M multiplied by the exit multiple. Inputs are stated above and are not benchmarks.
Exit multipleValue of +$1.1M net EBITDAWhat the buyer sees
8x$8.8MA conversion gain traced from source to signed contract
10x$11.0MThe same gain, valued at the example multiple
12x$13.2MThe same gain at a stronger multiple
  • Example only: $60M revenue, 15% EBITDA margin, $9.0M EBITDA, $20M of revenue from inbound leads.
  • Conversion 2.5% to 3.0%, which is +20% more customers from the same leads.
  • Added revenue $4.0M. At 35% flow-through, added EBITDA $1.4M.
  • AI and owner cost $0.3M a year. Net EBITDA +$1.1M, from $9.0M to $10.1M (+12.2%).

What is the 100-day AI plan for a portfolio company?

Three phases, 100 days, one number the board can read every Monday. The order is on purpose. Measure first, fix the handoff second, add AI where it pays third. Spend before the handoff is fixed buys faster leaks.

One rule runs through all three. Every AI step names a person who owns the result and a number it must move. KPMG's partners describe the shift in the same words: from vanity pilots to EBITDA discipline.

This plan matches the approach on our private equity portfolio page, which finds the leak, fixes the handoff and scales what sells. In one financial services operation we ran, cost per booked appointment fell 24% in two months (July to September 2026, see the case study). The method was measurement first.

The 100-day plan: baseline, response, then scale
  1. Pull 90 days of leads by source, first reply time, quotes open, wins and cost per customer. Time the first reply on the last 30 leads. Note data gaps. Pick the one number the board will read weekly: customers per 100 leads.

    ToolsCRM reports in HubSpot, Salesforce or HighLevel · Call tracking in CallRail or Invoca

  2. Follow ten real enquiries from first contact to signed or lost. Count every open quote and its owner. Rank the leaks by revenue. Choose the first AI target from the biggest leak, not from a vendor demo.

    ToolsCRM pipeline report · Call recordings in Invoca or Gong

  3. One owner per lead. An instant text and a call task on every new lead. A missed-call text-back. A next action and a date on every quote. AI drafts the follow-up and a person approves what goes to a client.

    ToolsWorkflows in HubSpot or HighLevel · Twilio for text · Zendesk or Intercom for first-line support answers

  4. Run every recorded call through call analysis. Find the three objections that lose jobs and the discounts that did not win. Coach on the calls that lost. Review price decisions against past wins.

    ToolsGong or Invoca for call analysis · Your CRM for quote and discount history

  5. Pick one cost line with clean data, such as indirect spend or support volume. Run one sprint. Write the proof file: baseline, change, date, owner and number. This is the file a buyer will ask for at exit.

    ToolsMicrosoft Copilot or ChatGPT Enterprise for analysis · A shared proof log in Notion or SharePoint

  6. Keep a step only if its number moved and a person owns it. Move budget toward sources that produce signed revenue. Add spend only where the path can carry it.

    ToolsWeekly board report

Which famous AI-in-private-equity numbers can you trace to a study?

Pages on this topic repeat a short list of big numbers. We traced seven. Six come with no sample, no named study or no method. One traces to a real data set, with limits.

The first is 84%. EY says 84% of PE funds expect AI to have a transformative impact and credits an unnamed recent EY study with no sample size or year. The second is an AI margin gain of over 10%. EY says this comes from its own experience, not a survey.

The third is 200 to 400 basis points of EBITDA expansion in 12 months, with a 0.5x to 1.5x multiple uplift at exit. A blog credits this to FTI's 2026 Private Equity AI Radar. The FTI report page we read shows 95%, 36%, 7%, 41% and 35%. It shows no basis point or multiple figure. Treat it as a best case, not a benchmark.

The fourth is half a turn to a full turn of extra EBITDA multiple for AI built across the hold. The article that states it credits "deal advisers" without a name. The fifth is that deals with proprietary AI carry multiples 15% to 20% above peers. The sentence carries no inline source, and the page's source list names a broker's valuation model with no sample. The sixth is BCG's 30% to 35% total return for AI on digital foundations. BCG gives it with no source or sample of its own, next to a reference to an IT buyers survey.

The seventh is real. As reported by Evident, McKinsey studied 471 PE-backed companies with $1M to $250M of revenue and found that median revenue multiples rise with AI depth: about 13x, 14x, 20x and 31x across four levels. McKinsey frames the finding as correlation, not cause. Stronger companies may simply adopt more AI.

Six of the seven famous AI-in-private-equity numbers have no sample or named study
  • 6 of 7
    No sample, no named study or no methodunnamed study, own experience, unsourced or unsized
  • 1 of 7
    Trace to a measured data setMcKinsey, 471 companies, correlation only

Megawebvision check of seven numbers repeated on AI in private equity pages, November 2026. Each row is in the table below.

Number you will seeWhat we foundVerdict
84% of PE funds expect AI to transform their businessAttributed to an unnamed recent EY study, no sample, no year (EY, 2025).Study not named
AI automation lifts margin by more than 10%EY states this comes from its own experience, not a survey (EY, 2025).Own experience
200 to 400 bps EBITDA and 0.5x to 1.5x multiple upliftA blog credits FTI (Wyecliff). The FTI report page lists no such figure (FTI, 2026).Untraced
Half a turn to a full turn of extra multipleCredited to unnamed deal advisers (HatchWorks).Untraced
Proprietary AI deals carry 15% to 20% higher multiplesNo inline source. The page's source list names an FE International valuation model, with no sample (HatchWorks).No sample
AI on digital foundations lifts total returns to 30% to 35%Sits beside an IT buyers survey reference, with no source or sample of its own (BCG, 2026).No source given
Median revenue multiple rises from 13x to 31x with AI depth471 PE-backed companies, correlation not cause (McKinsey, June 2026, as reported by Evident).Traced, with limits

AI in diligence: what will a buyer ask at exit?

A buyer's team will test your AI the way it tests your contracts. They will ask what is real, who owns it, what it costs and what happens if the key person leaves. Most of this is plain paperwork that a 100-day plan can produce as it goes.

The questions are consistent across the buyer guides we read. HatchWorks groups them in six dimensions: ownership, data, model dependency, team concentration, governance and modularity. Ontrac's buyer checklist adds a claims check: list every capability sold as AI and test it in production. It also asks what share of output ships with no human step.

Two facts raise the stakes. Bain's 2026 M&A report, a November 2025 survey of 303 M&A executives at companies above $100M of revenue, says one in five strategic dealmakers walked away from a deal because of the expected impact of AI on the target. Bain's PE report says the typical buyout exit now runs about seven years, against five to six in 2010 to 2021, so a hold has room for a proof file to build.

The premium for good AI evidence is not proven. Reported ranges vary and few name a study (see the table above). Plan for the downside: a company with clear records avoids a discount and a slow process. Any premium is a bonus.

  1. Inventory: every system sold or run as AI, with its owner, vendor and purpose.
  2. Proof of value: for each one, the baseline, the change, the date and the signed-off number.
  3. Ownership: signed IP assignments, vendor terms and change-of-control rights for models, code and prompts.
  4. Data rights: where the data came from, whether contracts allow its use, and how personal data is handled.
  5. Vendor dependency: share of gross margin spent on AI, the number of providers, the cost to switch.
  6. Key people: who can retrain, debug and run each system, with documents and retention in place.
  7. Governance: a usage policy, human-oversight records, an incident log and a risk class for each system.
  8. Claims check: no marketing copy or deck line that says more than the production data shows.

What should an operating partner check this week?

Start with one portfolio company and four counts from the last 90 days: leads by source, leads with a human reply inside five minutes, open quotes with an owner and a date, and customers. Everything else follows from these.

Do not buy a tool before you have the counts. A tool automates a process. If you do not know where leads stop, it automates the leak. Read the lead response time statistics for the first-reply data and the AI implementation roadmap for how to sequence the work.

Then ask three questions of the CEO. Which AI step has a named owner? Which has a number that moved? Which would we show a buyer today? Any step that fails all three is a pilot, not a lever.

The plannerExample: web form lead

Your follow-up plan, day by day

Pick the lead and the ways you can reach people. The plan below redraws. Nothing you pick leaves your browser.

Web form or chat

8 touches over 14 days, 4 of them calls.

A 2014 secret-shopper study found reps averaged 1.45 calls and 1.56 emails per lead.

4 questions, about a minute. No cost. Timur replies himself.

Questions leaders ask

How is private equity using AI?

Funds use AI in sourcing, diligence and monitoring, and push it into portfolio companies for revenue and cost. GPs most often name due diligence and deal sourcing as the highest return (Bain and StepStone, 103 GPs, 2026). In portfolios, 36% of companies use AI across use cases and 7% run it at enterprise scale (FTI, 2026).

Does AI add EBITDA in a private equity portfolio company?

It can, but proof is thin. 95% of funds say AI initiatives meet their business case (FTI, 2026), yet 39% of GPs expect no material financial impact in 2026 (Bain and StepStone). The gain comes where a company counts it: response, follow-up, pricing, support and spend. Count the baseline before you start.

How should a portfolio company start with AI in 100 days?

Measure first, then fix the lead and quote handoff, then add AI where a leak is largest. Days 1 to 30 set the baseline and find the leak. Days 31 to 60 speed up response and follow-up. Days 61 to 100 add call analysis and one cost sprint, and build the proof log.

Is revenue or cost the better first target for AI in a portfolio company?

Revenue first. Revenue gains compound and flow through to EBITDA at a high rate, and 41% of respondents name revenue acceleration as the top AI priority (FTI, 2026). Reported results still skew to cost savings (Bain and StepStone, 2026). Cost cuts come second, on one clean line at a time.

Does AI raise the exit multiple?

No study we found proves it. Claims of half a turn to a full turn, or 15% to 20% higher multiples, come with no sample or named study. McKinsey links deeper AI use to higher revenue multiples across 471 PE-backed companies, as correlation only. A clean proof file avoids a discount even if no premium appears.

What will a buyer ask about AI in diligence?

Buyers ask what is real, who owns it, where the data came from, how much it depends on one vendor, who can run it, and how it is governed. Buyer guides group this in six areas: ownership, data, model dependency, team concentration, governance and modularity. Bring a proof log and an inventory.

How much EBITDA growth does a deal need now?

About 10% to 12% a year for a 2.5x return over five years, against about 5% in the 2010s (Bain Global Private Equity Report, 2026). Higher borrowing costs of 8% to 9% and record entry prices leave less room for leverage and multiple expansion, so operating gains carry more of the return.

SOURCES

Where’s the money leaking?

Pick one. Be honest.

Cite this article

Grigorchuk, T. (2026, November 20). AI in Private Equity: The 100-Day Plan for Portfolio Companies. Megawebvision. https://megawebvision.com/insights/ai-in-private-equity

Text and infographics are licensed CC BY 4.0: reuse them with credit and a link to this page.