The seat has a name now. In February 2025, Korn Ferry published the paper that christened it: the “AI operating partner” – private equity’s newest value-creation role, already being carved out of the technology operating partner’s job at large firms. Heidrick & Struggles followed with a paper calling the role a new strategic imperative. When both major search firms describe the same seat within a year, the seat exists – at minimum as a line in someone’s budget.
Here is the problem, if you own portfolio value creation at a fund between $250 million and $2 billion. You have read some version of these papers, and you have probably run the hiring math – perhaps more than once, perhaps annually. And the math loses every time. That is true whether you are a dedicated operating partner watching the mandate fold into your generalist load, or the deal partner at one of the many funds this size – plausibly most – with no operating partner at all, where the mandate lands on you personally, stacked on top of the boards you already chair and the deals you are paid to find. No survey covers this bracket well enough to put a clean number on that split. The pattern is visible anyway.
So the seat is real, and the hire is not.
My argument is that this is not a stalemate. The same industry that invented the seat legitimized a second form of it in the same breath – Korn Ferry’s own words: AI operating partners “can be full-time employees or part-time advisors.” The question for a fund this size is not whether to fill the seat. It is which form of the seat to retain.
Before the argument, the obvious objection: is this not exactly what a senior AI advisor would say? It is. I sell the advisory form of this seat, so “the advisory form wins” is a conclusion I have an interest in reaching. For that reason I want to build the case out of the search firms’ own words and arithmetic you can check yourself – and if the arithmetic rescues the hire at your fund, you should hire.
The seat got invented
Korn Ferry describes large firms staffing the new role from three pools: entrepreneurial AI founders, technical product leaders, and executive technology leaders. Heidrick describes the mandate as portfolio-wide – deal sourcing and diligence support, operating work inside portfolio companies, fund-level reporting and risk – and describes the candidates bluntly: the profile is scarce, and the strong ones treat carry and portfolio-wide scope as must-haves. Heidrick also warns that funds which wait will be left hiring people with no private equity experience, with millions left on the table.
Search firms are in the business of describing seats that someone must then be paid to fill; papers like these are also marketing. That does not make them wrong.
Note one more detail in Heidrick’s description, because it matters later: the AI operating partner typically arrives with no direct reports or a very small team, and is expected to deliver measurable profit improvement anyway. Hold that thought.
The arithmetic below $2 billion
A fund in this bracket runs its management company on the fee stream – roughly 2 percent of committed capital, so somewhere between $5 million and $20 million a year across the bracket. That fee stream pays the salaries, the systems, and the raise-to-raise overhead of the firm itself. I am simplifying – fee step-downs, offsets, and recycled capital all complicate the picture – but not in a direction that rescues this hire.
Now price the candidate. Heidrick’s profile wants senior-executive cash compensation and, more to the point, carry. At a mega-fund, carry is deep enough that a new seat at the table is an accounting question. At a fund with a handful of partners, carry is the scarcest thing the firm has – it is how the people who founded the firm and source its deals get paid – and a non-investing seat that demands a slice of it is competing with the partnership itself.
Then divide by the portfolio. A firm holding several dozen companies can amortize a dedicated seat until it looks cheap. Across eight to fifteen companies – the normal shape of this bracket – the same seat costs more per company than the senior help it was supposed to replace. My own back-of-napkin threshold: somewhere north of a dozen portfolio companies, and often materially above that, the full-time hire stops penciling out. Below it, the seat never opens.
To be fair to the hiring case: at a large platform, Heidrick is right, and the funds announcing dedicated AI operating partners are right to hire them. The economics of the seat scale with the number of companies under it. Yours do not reach it. That is the whole disagreement – not whether the work matters, but whether the work at your scale is a full-time unit of anyone.
And where there is no operating partner at all, this arithmetic never even runs as a hiring question. It runs as a Tuesday question – answered somewhere between an investment-committee memo and a board deck, with the conclusion, once again, that the portfolio’s AI answer is still “each company is looking at it.” You know how that answer reads in an LP’s diligence questionnaire.
What the empty seat costs
The cost of leaving the mandate unowned is being measured now. AlixPartners’ 2026 leadership survey – the one major survey whose sample solidly includes funds under $5 billion – put it in a sentence: “smaller firms lag larger peers in AI adoption, creating a widening capability gap across the industry.” That sentence is about this bracket. It is the polite version of what limited partners are now probing for.
FTI Consulting’s 2026 survey of two hundred senior private equity respondents found that the top barrier to AI adoption is not budget, which ranked far down the list. It is the shortage of AI talent and skills – 35 percent put it first – and the leading responses were hiring (68 percent) and partnerships (51 percent). One caution, because precision matters with this number: FTI’s sample starts at $1 billion in assets under management. Those are funds a notch above much of this bracket, with more money and better access to the scarce profile. If they name talent as the binding constraint and reach for partnerships anyway, it is reasonable to expect the funds below them to be further behind and shorter on options – not better off.
The gap, in other words, is not a tooling gap. Tools are the one thing nobody in this market is short of. It is a seat gap – a missing owner of judgment – at funds that cannot pay what the seat now costs.
The advisory form of the same seat
Translate the mega-fund seat’s workload down to an eight-to-fifteen-company portfolio and something becomes obvious: the work is lumpy. A concentrated stretch of work at one company that is ready for it, at its leadership’s request. Diligence support when a deal has an AI-shaped question in it. A senior second opinion when a leadership team is deciding whether a proposed AI spend would earn its keep. Then stretches where the honest workload is nearly nothing. That is not a job. It is a fraction of a job that occasionally spikes – which is precisely the shape of work the advisory professions were invented for.
This is where Heidrick’s candidate spec becomes useful in a way Heidrick perhaps did not intend. One senior person. No team to speak of. Autonomous. Measured on profit outcomes rather than activity. Read cold, that is not a description of an employee at all – it is a nearly exact job description for a retained senior advisor. Heidrick set out to make the hiring case and, in passing, wrote the advisory form’s spec.
The early evidence points the same way, with a caveat I will state in full. MIT’s 2025 study of generative AI in business – a small early study its own authors call preliminary – found external partnerships reaching deployment at roughly twice the rate of internal builds, about 67 percent versus 33. The authors themselves caution that the difference may reflect organizational capability rather than anything causal, and I would not lean an investment decision on the number. But its direction agrees with what FTI’s respondents said they were already doing: when the constraint is scarce senior capability, buying access to it beats building it slowly.
There is also a dividend here that gets less attention, and it belongs to the portfolio companies. The advisory form is the only version of this trend under which a $20 to $150 million company ever gets the senior person. A full-time hire’s attention flows, rationally, toward the largest assets. A senior advisor retained across the portfolio shows up where the work is, works at the company leadership’s request, and leaves – without adding a permanent layer of oversight, and without a carry-priced salary that has to be justified onto the companies’ P&Ls. For the chief executive of a company that size, that is the difference between getting help and getting headquarters.
What retaining well looks like
Funds already know how to do this. At this size nobody puts transaction counsel on payroll, and nobody calls a quality-of-earnings firm a failed attempt to hire a CFO. Judgment that is needed in concentrated doses gets retained, not employed – that is the normal pattern of the business, and the AI seat is not an exception to it. The choice was never really hire-or-nothing. It is employed or advisory, and the buyer’s own literature has said so from the start.
If the advisory form is the right one, it can still be bought badly. Three tests I would hold any retained AI advisor to – including me:
First, senior and individual. The judgment you are retaining should belong to the person in the room – not to a leveraged team billing under a familiar name. Heidrick’s spec is the right spec: one senior person, autonomous, measured on outcomes. If the second meeting has more people in it than the first, you are buying a different product.
Second, the freedom to say “not here, not yet.” The most valuable sentence an AI advisor can produce for a fund this size is the one naming where AI should wait – where the data footing or the operating model means money would be spent to please the board rather than to move the numbers. Listen for whether the advisor can say it. A recommendation is worth exactly as much as the advisor’s freedom to recommend nothing.
Third, retained on the patterns you already use. Scoped work first. Standing advice only if the scoped work earns it. The same commercial shape as counsel and the diligence firms you already trust, applied to a newer kind of judgment.
The money follows the same patterns. Some funds will retain the advisory seat at fund level – standing counsel to the portfolio – and that door stays open. More often at this bracket, the split runs along a line the business already draws: the portfolio read costs the fund nothing, and the work it surfaces is engaged, owned, and paid for by the company it serves – the same arrangement as every other piece of post-close operational help your companies buy. That keeps the seat off a fee stream the arithmetic above showed was spoken for, and it means nothing is owed until a company decides a specific piece of work is worth it.
I should say where my conviction about this motion comes from, because it is firsthand. I spent years as VP of analytics inside a roughly $1 billion-revenue, PE-owned telecom, and I arrived there exactly the way I am describing – an outside expert introduced through the sponsor’s operating partner, initially for a scoped piece of work, brought in because the leadership wanted help it could not get from inside. (The engagement went well enough that the company eventually made the seat full-time, which I concede is the employed form winning a round.) I have watched what this motion does inside a portfolio company when the work is done at the leadership’s request rather than over its head. The motion works. That experience, more than any survey, is why I think the advisory form is not a consolation prize for funds that cannot afford the hire. At this scale, it is simply the correct form of the seat.
The upshot fits in a sentence: retain the judgment; keep the capability. The judgment comes from outside, in the doses the portfolio needs. The capability – the working habits and the first deployed use cases – stays inside the companies, with the teams who will still be there after the advisor leaves.
The bounded first step
Budget season will come around again, and with it the same conversation: should we hire an AI operating partner this year. The honest answer, at this bracket, will still be no. The harder follow-up is what the fund is doing instead, because “nothing, again” is now a visible answer – LPs can see it in a diligence questionnaire, and exit buyers could soon be pricing it. Which form of the seat to retain is a judgment call. Whether the seat stays empty another year is the question I would not want to be answering in front of either audience.
The rational way to test the advisory form is a bounded sample of it. Mine is called The Portfolio AI Baseline: a fund-level read of where AI can move the numbers across your portfolio, and where it should wait. It costs the partner under three hours – one working session and a playback – with the data request handled by an analyst on your team and the desk work happening off your calendar. It is free for a commercial reason, not a charitable one: the baseline is how I choose the work, and I make money only if the fund decides to continue past it. The baseline is allowed to conclude that you should not. What the first piece of work should be, and why, is its own discipline – I take that up in a companion piece.