An AI consulting contract for a portfolio company should name the system built inside the company, its owner, and the test it must pass.
In this post, a portfolio company is a company that a private equity fund owns. An operating partner is the person at the fund who works with those companies to improve their results. AI consulting is outside help with artificial intelligence, from advice and slide decks to software built and run in the company. Naming the system, its owner and its test shows whether the company is buying advice or a working system.
The system is the AI software the consultant builds. It should run in the portfolio company's own software accounts, on the company's own data. The portfolio company should sign the contract and pay for that software, so the system stays with the company when the fund sells the company. The contract should also say who pays the consultant's fee: the fund, the company, or each a stated share.
The owner should be one manager in the portfolio company whose budget pays for the system, such as the head of operations. At handover, the consultant should give the owner administrator access, meaning the right to change the system. The consultant should also hand over the written settings and the test results.
A Federal Reserve note from April 2026 cites early workplace surveys of generative AI, the kind that drafts text. The note dates those surveys to 2023 and 2024. They found a "non-trivial amount" of AI use among employees occurring "without the knowledge or approval of IT or management". So the consultant's first piece of work should be a list of the AI tools staff already use. The list should include agents, programs that carry out a task without a person approving each step.
A working paper from the National Bureau of Economic Research, dated March 2026, surveyed nearly 750 corporate executives. Most of them were chief financial officers. The paper has not been peer reviewed.
On average, the executives said AI raised output per worker by 1.8 percent in 2025. The executives also estimated their revenue and their headcount. Those two estimates together implied a smaller gain per worker, 0.6 percent. The authors write that the gap is likely "reflecting a delay in revenue realizations", meaning sales that arrive later.
So the test should measure cost or time on one task the company already tracks. Any change in revenue should be reported separately. The test should name the task, its current cost or time, the target, and the number of weeks it runs. The owner decides at the end of those weeks whether the system stays.
The National Bureau of Economic Research paper also reports a separate survey, with 183 valid responses, on the type of AI spending. Firms with 500 or more employees in that survey expected, on average, 55 percent of 2026 AI spending to go on running costs. The paper lists those costs as subscriptions, services and training. The same firms expected 31 percent to go on building or customizing their own AI systems. The contract should place each consulting fee under running costs or under building, so finance can see which one the company is paying for.
What to do first:
- Operating partner: before any AI consulting contract is signed, have it name the system, its owner, its test and who pays the consultant's fee.
- Portfolio company chief financial officer: before work starts, pick one task finance already tracks and record its current cost or time.
- Portfolio company head of operations: in the first month, have the consultant list the AI tools and agents staff already use, approved or not.
Bain & Company, a consultancy, ran a survey with StepStone Group, a private markets investment firm. It asked more than 100 investment and investor relations professionals, mainly in North America and Europe, between December 2025 and January 2026. The survey asked how general partners, the firms that run buyout funds, are approaching 2026. Bain and StepStone's press release of March 2, 2026 says AI benefits inside portfolio companies skew toward cost savings. Nearly 40% of general partners did not expect material financial impact from AI in 2026.
