Turn the Board's AI Mandate
Into a Number.
The AI Operating Diagnostic takes one portfolio company end to end: where AI moves margin, what to build first and what it is worth annualized. It ends with a figure the CFO can defend, a ranked build order and one working workflow running in the business.
Run personally by a fractional Chief AI Officer and 3x CHRO with 11 AI builds shipped, and a Contract AI Model Trainer for OpenAI, Meta and Microsoft.
The Problem
The Mandate Is Real.
The P&L Impact Is Not.
Almost every portfolio company now has an AI mandate from the board. Almost none can point to the line it moved. Portfolio AI does not fail loudly: it stalls in one of six specific, recognizable ways, and every one of them is an operating failure rather than a technology failure.
The License Trap
You pay for AI seats (Copilot, ChatGPT Enterprise) but cannot show usage or output that maps to a dollar.
The Strategy Shelf
There is an AI strategy deck. Months later, nothing has actually shipped from it into the business.
The Demo Graveyard
Pilots demo well, then die before production. Curated data works; the real systems and workflows do not.
The Workshop Certificate
The team completed AI training. Day-to-day behavior and output did not change afterward.
Vendor Lock-in
The AI lives inside a vendor tool you do not control or own. Leverage and IP sit outside the company.
No Adoption Owner
No single owner, no incentives and no change plan for getting the workforce to actually use it.
Of companies that measured AI cost savings landed below 10%, far short of the returns they expected
Bain & Company
What most portfolio companies can point to on the P&L today, a year into the mandate
"Companies do not fail at AI because the tools do not work. They fail because they build before they know what is actually worth building."
Yuri Kruman
Score one portfolio company: 3 minutes, free →The Scorecard is the entry point. The result sets the scope of the Diagnostic.
The Commitment
We Map 3x the Fee in Annualized Margin
and Ship One Working Workflow.
Or you keep the prototype. Both commitments go into the engagement letter before anything starts.
3x the fee, annualized
The value model is built on the portfolio company's own loaded cost and volumes, not on industry averages, so your CFO can check the arithmetic line by line. If we cannot map at least three times the fee in annualized margin, we have not done the job.
One working workflow, shipped
Not a slide about a workflow. One real thing running against your real systems, so the highest-ranked recommendation is demonstrated rather than asserted. The prototype is yours to keep either way.
Credits in full toward any retainer or build
If you go forward with a Fractional Chief AI Officer retainer or a build, the entire fee credits toward it. There is no pressure to go forward: the Operating Map, the value model and the prototype are yours regardless.
A Do-Not-Automate list, every time
Every Diagnostic names where automation would waste money and where human judgment is the actual value. An independent diagnosis has to be able to tell you to spend nothing, and this one does.
Terms in plain English in our Terms.
What It Produces
The Operating Map
Eight outputs. One defensible answer to the question the board keeps asking. Yours to keep.
Margin Map
Every function in scope mapped to the line it actually touches: cost of goods, SG&A, working capital, revenue per head. Each step scored on volume, how much is rule-based versus human judgment, data readiness and adoption risk. You see where the margin is, not where the buzz is.
The Annualized Value Model
The number, annualized and defensible, built at the company's real loaded cost and real volumes. This is the figure the 3x commitment is measured against, itemized so the CFO can pull any line apart and rebuild it.
Stall Diagnosis
Which of the six stall modes are live in this portfolio company, scored, with the specific operating cause of each. The free Scorecard gives you the self-assessed version. This is the evidenced version, taken from the workflows themselves.
Build, Buy or Stop
Every opportunity assigned: where a vendor is genuinely the right answer, where a custom build keeps the leverage and the IP inside the company and where the honest answer is to stop paying for something you already have. Each with an estimated build cost and payback.
The Do-Not-Automate List
Just as important as what to build: where human judgment is the actual value, so the company does not burn budget and credibility automating the wrong things. This is what an independent diagnosis looks like.
Adoption Plan and Named Owner
Who owns adoption inside the company, what they are measured on and what changes in the week-to-week so the workforce actually uses what gets shipped. Most portfolio AI dies here, so this is written as an operating plan, not a change-management theme.
One Working Workflow
The top-ranked recommendation, built and running against your real systems rather than a curated demo set. It is the difference between a thesis and a proof, and it is the thing that keeps this out of the Demo Graveyard. Yours to keep.
IC-Ready Brief and Live Readout
One page for the investment committee and the fund's operating partner: the number, the top three moves, the build order and what each is worth. Delivered in a live readout with the management team, recorded so the people who were not in the room still get it first-hand.
How It Runs
One Portfolio Company, End to End, in 30 Days
Five phases across 30 days, run personally, scoped to the functions where the margin actually sits. Long enough to map the margin, model the annualized value and ship one working workflow into the business, not just a deck.
Scope and Access
A working session with the CEO or COO to set the perimeter: which functions are in scope, what the board is actually asking for and what has already been tried. NDA and read-only access are agreed here, before anything is shared.
Function-by-Function Scan
Every workflow, tool and cost center in scope gets mapped and scored for what AI should own, what it should assist and what it should never touch. This is where the six stall modes get diagnosed from evidence rather than from a survey.
Quantify and Rank
The value model is built against the company's own loaded cost and volumes, then validated with the CFO or finance lead so nobody is arguing about the inputs later. Every opportunity is ranked by margin impact against implementation difficulty.
Build One Thing
The top-ranked workflow gets built and run against the real systems. This is the part a consulting assessment never does, and it is the reason the recommendation survives contact with the business.
Readout and Build Order
A live walkthrough of the Operating Map and the build order with the management team and, where the fund wants it, the operating partner. Recorded and paired with the one-page IC brief.
Who Runs It
The Person Doing the Diagnosis
Is the Person Who Builds the Fix.
Most portfolio AI assessments are run by people who have never shipped the thing they are recommending. This one is run personally by a fractional Chief AI Officer and 3x CHRO who has shipped 11 AI builds for real organizations, coached 2,300+ executives, and is a Contract AI Model Trainer for OpenAI, Meta and Microsoft. The margin case is credible because the builds are real.
DueDrill · PE/VC Fund
40 to 60 hrs → under 4 hrs per deal
An AI due-diligence platform that tripled the fund's deal-evaluation capacity without adding headcount.
AI HR Pilot · HR Operations
3 days → under 2 hrs per ticket
An HR policy assistant that became a commercial product, ROI positive within 30 days of first deployment.
ChaiRaise · Nonprofit Fundraising
10+ hrs/week saved on outreach
An AI-native donor CRM that took fundraising from zero to a full pipeline and a first HNW commitment in 3 weeks.
Ravenbridge · Institutional Funds
$15K delivered vs a $120K agency quote
Three institutional fund websites delivered in 3 days, at a fraction of agency cost. Read the breakdown →
Scope the Diagnostic
One Fee. One Number. One Thing Shipped.
Starting fee, fixed. One portfolio company, end to end.
- check_circleThe full 8-part Operating Map, yours to keep
- check_circle3x the fee mapped in annualized margin, at your own loaded cost
- check_circleOne working workflow shipped, or you keep the prototype
- check_circleRanked build order plus the Do-Not-Automate list
- check_circleOne-page IC brief and a recorded live readout
- check_circleCredits in full toward any retainer or build
Fifteen minutes to confirm fit, scope and the functions worth going after. If it is not a fit, I will say so on that call and point you somewhere better.
Not there yet? Score one portfolio company free in 3 minutes and see which stall mode to fix first.
Against a stalled mandate
A mandate that stalls does not cost nothing. It costs the license spend, the executive attention and a stretch of the hold period you do not get back, and it ends with the same question unanswered at the next board meeting.
Against a $400K hire
An internal AI leader is roughly a $400,000 commitment, plus a search, plus a ramp, before the first system ships. The Diagnostic answers the question first, so the hire is made against evidence rather than against a hunch.
Questions
Straight Answers
What is the AI Operating Diagnostic?expand_more
A fixed-fee engagement that takes one PE portfolio company end to end. It maps where AI moves margin function by function, scores which of the six stall modes are live, converts the opportunity into an annualized number at the company's own loaded cost, ranks what to build first and ships one working workflow so the answer is demonstrated rather than asserted. You get the full Operating Map, an IC-ready brief and a live readout. It starts at $18,500 and credits in full toward any retainer or build.
What exactly is the guarantee?expand_more
We map 3x the fee in annualized margin and ship one working workflow, or you keep the prototype. Both commitments are written into the engagement letter before anything starts. The margin figure is built on the portfolio company's own loaded cost and volumes, not on industry averages, so the CFO can check the arithmetic line by line.
Why does it cost $18,500?expand_more
Because the alternatives cost more. A stalled AI mandate burns license spend, executive attention and a hold period you do not get back. A full-time internal AI leader is roughly a $400,000 commitment before the first system ships. The Diagnostic is a fixed fee that ends with a number, a ranked build order and one working workflow, and every dollar of it credits toward the retainer or build if you go forward.
Do we have to retain you afterward?expand_more
No. You own the Operating Map, the value model, the build order and the working workflow, and you are free to execute in house or with any provider. If you do go forward with Portfolio Leverage Co., the fee credits in full toward any retainer or build.
Where does the free Scorecard fit?expand_more
The AI Pilot-to-P&L Scorecard is the entry point and it is free. It takes three minutes, scores one portfolio company across the six ways AI stalls before the P&L and tells you which failure mode to fix first. Most Diagnostic engagements start there, because the Scorecard result sets the scope of what the Diagnostic goes after.
Who needs to be involved from the portfolio company?expand_more
The CEO or COO owns the engagement, the CFO or finance lead validates the loaded-cost assumptions behind the value model and the function leads whose workflows are in scope give a walkthrough of how work actually runs. The fund's operating partner is welcome in the readout and usually is, since the output is written for the investment committee as well as the management team.
How is this different from a consulting firm's AI assessment?expand_more
An assessment ends in a deck. This ends in a number your CFO can defend, a ranked build order and one workflow running in the business. The person doing the diagnosis is the person who builds the fix: a fractional Chief AI Officer and 3x CHRO with 11 AI builds shipped. Every engagement also includes a Do-Not-Automate list, which is where we tell you to spend nothing.
What do you need from the company?expand_more
A kickoff working session, read-only visibility into how work actually flows through the functions in scope plus loaded-cost and volume data for those functions so the value model is built on real inputs. Access is scoped and covered by an NDA before anything is shared.
Stop Reporting Activity. Report the Number.
One portfolio company, end to end: where AI moves margin, what to build first and what it is worth annualized. We map 3x the fee in annualized margin and ship one working workflow, or you keep the prototype.
Book the Scoping Call →From $18,500 · Credits in full toward any retainer or build · One portfolio company, end to end
Want the free version first? Take the 3-minute Portco Scorecard →