Private equity and venture fund
DueDrill
AI due-diligence platform for a PE/VC fund (defense tech, AI, health tech).
40 to 60 analyst hours per deal to under 4; deal-evaluation capacity tripled without added headcount
For operating partners and portfolio-company leadership
A fractional AI Operating Partner who ships the system and owns adoption.
For funds with fewer than about fifteen portfolio companies, and for the portcos they own between $20M and $500M in revenue, carrying a board-level AI mandate with no number against it yet. The work is measured in basis points of EBITDA inside the hold period. It is run by a practising 3x CHRO with a JD who builds the systems personally, so the change plan and the code arrive together.
3x Chief Human Resources Officer
JD, Benjamin N. Cardozo School of Law
#6 Global HR Thought Leader, Thinkers360 (2025 and 2026)
Contract AI model trainer for OpenAI, Meta and Microsoft
The state of portfolio AI
Why the mandate stalls
The sponsor asked for AI. The deck exists. The licenses are paid for. What is missing between that and the P&L is a person who owns the outcome inside the business.
36%
of PE-backed portfolio companies use AI in day-to-day operations. 7% call it fully integrated across the portfolio.
9%
of operating partners have seen a demonstrable AI premium in a completed transaction.
68%
of PE-backed CFOs do not know where to begin, while 98% of sponsors have told them to prioritize AI.
The bottleneck is adoption, not the model
Every portfolio company already has access to the same frontier models at the same price. That is not where the difference is made. Talent and capability is the primary constraint on portfolio AI, cited by 35% of respondents. In practice that means nobody senior is accountable for whether the tool is used, the workflow around it is unchanged, and the training was a one-hour webinar nine months ago.
This is an organizational failure with a technical surface. It is fixed by naming an owner, redesigning the work, retraining the people who do it, and putting the result in the same monthly report as everything else the board reads. That is the job this seat does.
More than 80% of AI programs still fail, usually due to misalignment on use cases, user adoption and unclear success measures.
Not one of those three failure causes is a technology problem. All three are ownership problems.
The seat
What a fractional AI Operating Partner actually does
Six responsibilities. All six are held by one person, and that person is accountable to the sponsor and to the portfolio company's board at the same time.
- Owns the AI line of the value-creation plan
- The line stops being a heading and becomes a scoped initiative with a target, a timeline and a named owner inside the business.
- Runs a 100-day plan per function
- One function at a time, sequenced by impact against ease. Finance, HR, revenue operations, service delivery, compliance. Not all of them at once.
- Ships the working system
- Built and put into production, not a vendor pilot with a renewal date. If a bought tool is the right answer, we say so and implement it properly instead of building for the sake of it.
- Designs adoption
- A named owner, incentives that match the target, role-based training, workforce design and the compliance posture that a JD and 3x CHRO brings to the same table as the build.
- Reports monthly to the board
- In a number the CFO recognizes, in the CFO's own format. Baseline, delta, adoption rate, what is next. One page.
- Builds the playbook
- So the second portfolio company costs less than the first, and the fifth costs less than the second. The asset the fund keeps is the repeatable rollout.
Against the alternatives
| Option | What it does | Where it stops | Cost |
|---|---|---|---|
| Strategy boutique | Advises. Frameworks, use-case inventory, a roadmap. | Nothing is built and nobody inside owns the rollout. | Six figures, quoted |
| Development shop | Builds. Ships an application against a specification. | Adoption, training and the P&L number are your problem. | Per project |
| Full-time AI operating partner | Owns everything, permanently, on the fund's payroll. | Only pencils above roughly fifteen portfolio companies. | $400K to $700K all-in |
| Fractional AI Operating Partner | Owns adoption and ships the system personally. | Two days a week, so sequencing matters. One function at a time. | $19,650 per month |
Two-thirds of operating partners already cover five or more portfolio companies, which is why the AI mandate lands on a bench that is already full. Not Very Private Equity, 2026
The method
How it runs: Diagnose, Own, Ship, Measure, Repeat
Five verbs, in order, per function. The sequence is the product. Skipping Own is why most programs end at Ship.
-
01
Diagnose
Two weeks inside one portfolio company. Workflows, loaded labor cost, and the three things not to automate.
-
02
Own
One named owner per function, written into the value-creation plan with a target against their name.
-
03
Ship
The working system in production inside 100 days, with the workflow around it redesigned and the team retrained.
-
04
Measure
Baseline before, delta after, adoption rate alongside. Monthly, in the format the board already reads.
-
05
Repeat
The playbook moves to the next function, then the next portfolio company, at a lower cost than the first.
Average holding periods are at a record 6.6 years, so a program that takes two years to show a number has spent a third of the hold. McKinsey Global Private Markets Report 2026
The ladder
Six ways in, priced
Start anywhere. Most funds start at the diagnostic, most portfolio companies start at the scorecard, and the retainer is where the seat is actually held.
-
AI Pilot-to-P&L Scorecard
Free
3 minutes · For any portco leader or operating partner who wants a read before a conversation
- Six failure modes scored 0 to 3 for one portfolio company
- A band: On Track, Pilot Purgatory or Acute
- The fix for the top mode, emailed back
-
Portco AI Diagnostic
$7,500
2 weeks, one portfolio company · Credited against a sprint or retainer within 90 days
- Workflow and adoption map for the functions in scope
- EBITDA-impact model in the portco's own loaded labor cost
- Opportunity matrix ranked by impact against ease, plus the do-not-automate list
- 100-day plan and a board one-pager
- A 60-minute recorded readout with leadership and the sponsor
-
100-Day Pilot-to-P&L Sprint
From $45,000
100 days, one function · For portfolio-company leadership with a mandate and a deadline
- One function, one accountable owner, one shipped system, one number
- Adoption design: incentives, workflow redesign, role-based training
- Workforce and compliance review of what changes for the people doing the work
- Baseline set in week one and a measured delta at day 100
- Board report and a written handover so the system runs without me
-
Most funds land here
Fractional AI Operating Partner
$19,650 / month
6-month minimum · About two days a week equivalent · Fund or portfolio company
- Owns the AI roadmap and the sequencing across functions
- Owns adoption: named owners, incentives, training, workforce design
- Owns governance: policy, data handling, model use and the compliance posture
- Owns the build backlog and ships from it personally
- Monthly board reporting in basis points of EBITDA
Portfolio pricing on request for funds running the seat across more than one portfolio company.
Book a portfolio call -
Portfolio Playbook
Quoted, declining
Per rollout · For operating partners repeating a win across the portfolio
- The sprint's system and change plan re-deployed at the next portfolio company
- Per-company fee declines with each rollout as the playbook hardens
- Standard adoption pack: owner brief, training, incentive design, board template
- Fund-level rollup so the operating partner sees every portco on one page
-
Exit Evidence Pack
Quoted
3 to 4 weeks · For funds heading into a process inside 24 months
- Adoption documented by system, by function and by rate
- Named ownership and the operating cadence that keeps it running post-close
- EBITDA attribution with the working and the assumptions exposed
- Answers written for the questions a buyer's diligence team will ask
Proof
What shipped and adopted looks like
Three systems built for organizations that had the same mandate. Each one is in production and each one is used by the people it was built for.
Venture studio portfolio
AI HR Pilot
HR policy engine for a venture studio's portfolio companies.
Ticket resolution from 3 days to under 2 hours; ROI positive within 30 days; became a commercial product
Multi-entity workforce
i9Drill
I-9 and work-authorization compliance auditor across a workforce.
Turns I-9 compliance into a standing audit-ready posture across every operating company
EBITDA impact model
Size the number before you buy the seat
One function, one shipped system, one adoption rate. Move the inputs to your portfolio company and see what the recovered cost is worth in basis points of margin and in enterprise value at your exit multiple.
$20M$500M
5%40%
5200
$60K$200K
5%40%
20%90%
6.0x14.0x
Annual recoverable cost
$532,000
Basis points of EBITDA margin
66 bps
EBITDA today about $12.0M. The recovered cost is about 4.4% of it.
Implied enterprise value at 9.0x
$4.79M
Retainer months covered
27
Months of the $19,650 retainer paid for by one year of recovered cost.
This is an order-of-magnitude model, not a forecast. It sizes one function under one set of assumptions. The diagnostic replaces every input here with the portfolio company's own numbers.
Show the arithmetic
recoverable = employees × loaded cost × recoverable share × adoption rate
basis points = recoverable ÷ revenue × 10,000
enterprise value = recoverable × exit multiple
retainer months = recoverable ÷ 19,650
Recovered cost is treated as flowing straight to EBITDA, which is the optimistic case: in practice some of it is reinvested as capacity rather than taken out as cost. Basis points are expressed against revenue, which is the same thing as the points added to the EBITDA margin. The enterprise-value line assumes the multiple holds and the saving is durable enough for a buyer to underwrite it.
Operating rules
How the seat behaves once it is in the building.
Written down so a sponsor can hold the engagement to them.
Instrument first, intervene second.
The baseline is measured in the portco's own loaded labor cost before anything ships. No build starts on an estimate.
Every build gets a number the board already reads.
Recoverable cost in basis points of EBITDA, adoption rate by week. Never seats licensed or people trained.
The portco's finance lead counts the result, not PortLev.
The attribution that reaches the board is theirs. The exit evidence pack is built from their ledger.
One accountable operator, named collaborators.
Yuri Kruman leads every engagement personally. Where a build needs a specialist (security review, data engineering), that person is named in the statement of work. There is no anonymous bench.
Leave the company able to run it.
Three to five trained operators, the portco's own IP, and a written handover. The engagement ends when the team no longer needs the seat for that function.
Fit
Who this is for, and who it is not for
This is for you if
- You run a fund with fewer than about fifteen portfolio companies and no full-time AI operating partner.
- You lead a portfolio company between $20M and $500M in revenue with an AI line in the value-creation plan and no number against it.
- You will name an owner inside the business and give them a target.
- You want the result in the CFO's reporting, not on a vendor dashboard.
- You have an exit inside 24 months and buyers will ask what AI actually does here.
This is not for you if
- You want a strategy deck and an inventory of use cases with nothing built.
- The point of the pilot is to demonstrate innovation rather than move a line in the P&L.
- No executive inside the business will own adoption.
- You are comparing seat licenses on price and want the cheapest.
- The company is pre-revenue or has no operating cadence to plug into yet.
Questions operating partners ask
What is a fractional AI operating partner?
It is the operating-partner seat for AI, held part time. The role owns the AI line of the value-creation plan for one portfolio company or several: the roadmap, the build backlog, adoption, governance and the number that goes to the board each month. Large funds hire the seat full time. Below roughly fifteen portfolio companies the arithmetic favors the fractional form, because the work is a two-day-a-week job that nobody currently holds.
How is this different from a fractional Chief AI Officer?
It is the same seat viewed from a different chair. A Chief AI Officer sits inside one company and reports to its CEO. An AI Operating Partner sits alongside the sponsor and is measured on value creation inside the hold period, which scopes the work to what a buyer will pay for at exit. PortLev holds either title, and the deliverables, cadence and fee are identical.
How does the fee compare with a full-time hire?
A full-time AI operating partner runs roughly $400,000 to $700,000 all-in once bonus and carry are counted. The retainer is $19,650 per month, so the six-month minimum is $117,900 and a full year is $235,800 for about two days a week. The full-time hire only pencils above roughly fifteen portfolio companies, and two-thirds of operating partners already cover five or more. Below that line the fractional seat buys the same ownership without the fixed cost.
What does a 100-day sprint deliver?
One function, one accountable owner inside the business, one system running in production and one number the CFO already reports. Adoption design, training and the board report are part of the sprint, not add-ons. It starts from $45,000 and ends with a measured baseline, a measured delta and a written handover so the system runs without me.
How do you measure impact?
In basis points of EBITDA margin, calculated on the portfolio company's own loaded labor cost and reported in the CFO's own format. We set the baseline in the first two weeks, then report the delta monthly against it. Hours recovered, tickets closed and adoption rates are tracked, but they are inputs. The number that counts is the one already in the board pack.
What does the portco team need to provide?
A named executive sponsor, a function owner who will carry the adoption target, and three to five people for ninety-minute interviews. Read-only access to the systems in scope, plus whatever the CFO uses for loaded labor cost. After intake that is about two hours a week from the sponsor. If nobody inside the business will own adoption, the engagement should not start.
What happens at exit?
AI reaches diligence as a durability question: who runs it, what it saves, and whether it survives a change of ownership. The Exit Evidence Pack documents adoption rates, named owners, the systems in production and the EBITDA attribution in three to four weeks, in the format a buyer's diligence team expects. With average holding periods at a record 6.6 years, the evidence has to hold up long after the build.
Put a number against the AI line before the next board meeting.
Thirty minutes on one portfolio company. Where the mandate is stuck, which function pays back first, and what the first hundred days would cost.