Most buyout firms built their operating partner bench on a straightforward premise. Portfolio companies need help that their management teams cannot supply alone, and the fund can supply it faster and more cheaply than the open market. Over two decades that premise hardened into headcount. A mid-market fund with a dozen portfolio companies now routinely carries operating partners, sector advisers, a value creation team and the reporting apparatus that surrounds them, and charges a portion of it to the funds.
We think that structure is the first part of the private equity operating model that capable AI will reprice, and that it will move faster than diligence or fund finance, both of which currently get far more attention.
The reason is uncomfortable to say plainly. A large share of what value creation teams actually produce is diagnosis, benchmarking and playbook transfer. All three are forms of applied pattern recognition over documents and data. All three compress.
Most of the value creation function is reading, comparing and writing
Watch what happens in the first hundred days after a deal closes. Someone reads the diligence file again, properly this time, because the deal team read it for risk and the operating team needs it for opportunity. Someone rebuilds the management accounts into a shape the fund recognises. Someone baselines a set of operating metrics the company has never formally tracked. Someone compares the pricing structure against two other assets the firm has owned in adjacent sectors. Someone works out that procurement is fragmented across four legal entities and writes the memo that says so. Someone assembles the hundred day plan, circulates it for comment, and stands up the monthly reporting pack.
That is skilled work and it produces real value. It is also, at its core, reading, comparing and writing. A capable model with access to the fund's document estate and the company's systems does the reading in hours rather than weeks, runs the comparison against every asset the firm has ever owned rather than the two the operating partner happens to remember, and drafts the memo to a standard a senior person can edit rather than originate.
The output is not better than a good operating partner's first draft. It is roughly as good, available immediately, and available for every asset at once rather than for the two or three currently receiving attention. That last point is what reprices the function.
Benchmarking was valuable precisely because it was inaccessible
An experienced operating partner's benchmark is carried in their head. They know what gross margin ought to look like in a specialty distribution business at forty million of revenue because they have run three of them. They know the finance function is understaffed before they have seen the organisation chart. That knowledge was scarce, it took twenty years to acquire, and firms paid handsomely for it because there was no other route to it.
The scarcity, though, was never in the knowledge. It was in the retrieval. Firms have always sat on enormous quantities of comparable material: every management pack from every company they have owned, every commercial diligence report, every exit memo setting out what worked and what did not. Almost none of it was reachable. It sat in dormant deal folders, in a data room whose licence expired, or in the head of a partner who left in 2019.
When that corpus becomes readable, the benchmark stops being a personal asset and becomes an institutional one. A firm can ask what happened to working capital in the four months following every ERP migration it has financed, and get an answer grounded in its own history rather than a recollection of it. That is a genuine improvement in quality. It also removes the main reason a particular individual was indispensable.
The best operating partners were never doing the commoditised part
Here is the strongest counter-argument, and we think it is correct. Ask a general partner to name their best operating partner, then ask what that person actually does. The answer is almost never diagnosis. It is that they persuaded a founder to hire a real chief financial officer. It is that they sat with a management team through a covenant breach and stopped them panicking. It is that they knew which of two plausible turnaround candidates would survive the culture, and they were right.
None of that is pattern recognition over documents. It is judgement applied under pressure by someone with enough standing that a chief executive will take the advice. Our own view is that the people at the top of this profession have always been in the intervention business, and that the analysis was the ticket of entry rather than the product.
So the argument is not that the function disappears. It is that it loses the layer beneath its best people, the layer that existed to produce the analysis those people used as an opening position. That layer is precisely where the headcount sits.
Presence, credibility and the willingness to be disliked do not compress
Consider what it takes to change a management team. The analysis showing that the chief operating officer is not equal to the next stage of growth is not the difficult part, and it is usually obvious to everyone well before anyone writes it down. The difficult part is having the conversation, managing the founder's reaction, finding the replacement, holding the business steady through the gap, and carrying the consequences if the judgement was wrong.
The same holds for most real interventions. Renegotiating a critical supplier contract needs someone in the room whom the supplier believes. Getting a hundred day plan executed needs a person management will return calls to in week eleven, when the plan has become inconvenient. These are accountability and relationship functions, and they are not improved by being made faster.
What does improve is the preparation behind them. An operating partner who walks into a monthly review already knowing which three of forty variances are worth discussing arrives in a materially stronger position than one who spends the first hour of the meeting establishing that.
The arithmetic moves before the organisation chart does
To put rough numbers on this, and these are illustrative rather than measured, take a mid-market fund with fourteen portfolio companies and six people in the value creation function, two of them genuinely senior. Assume the six cost the firm and its funds somewhere in the region of four million dollars a year fully loaded. Assume, again illustratively, that around half the team's time goes on diagnostic, benchmarking and reporting work rather than on intervention, which is roughly what we see when firms trouble to log it.
On those modelled assumptions, something like two million dollars a year is being spent on work whose marginal cost is falling towards the cost of running a model across the documents. The firm will not capture that as a saving in year one, because the people are employed and because coverage today is thin rather than generous. What it captures first is reach: the same six people covering fourteen companies seriously instead of the five that were getting real attention.
That is why we expect the repricing to appear as a change in what the function is asked to do well before it appears as a change in its size. Management fee budgets are sticky, partnership agreements are stickier, and firms rarely shrink a team they have just learned to use differently. The compression arrives at the next fund, when the team is designed rather than inherited.
Fewer people, more senior, closer to the company
The version of this we find most plausible is not a smaller value creation function delivering less. It is a differently shaped one. The analytical base narrows, because the analysis is largely produced rather than performed. The senior layer holds or grows, because intervention capacity becomes the binding constraint once diagnosis stops being rationed. And the coverage model changes, so that every portfolio company receives a proper diagnostic rather than only those large enough or troubled enough to justify the effort.
There is a second-order effect worth naming. If diagnosis is cheap and continuous, problems surface earlier. Much of the permanent damage in underperforming assets comes from the gap between the moment a trend becomes visible in the data and the moment somebody senior looks at it. Closing that gap is worth considerably more than any saving on the analytical layer, and it is the argument we would put to an investment committee rather than the cost one.
For the individuals concerned this is a better job, not a worse one. The part of the operating partner role that people describe as a grind is the part that is compressing. What remains is the part that people who are good at it describe as the reason they took the job: sitting with a business that is not working, deciding what to do about it, and staying there while it gets done. Fewer of them, better used, and pointed at the companies that need them rather than the ones whose reporting pack happened to arrive first.
