The middle-market PE world is loudly buying into AI at the holding level and quietly failing to deploy it at the portco level. Both things are true at the same time.
The holding-level signal: every mid-market PE fund we have talked to in the last six months has either hired or is recruiting some version of an AI Officer. Operating Partner with AI in the title. Chief Innovation Officer with AI as the explicit mandate. Group-level 'AI Center of Excellence' billed to portcos as a shared service. The rhetoric is consistent and the budgets are real.
The portco-level reality: most portfolio companies still cannot consolidate their multi-business operating cadence on the same calendar. Month-end close runs differently across portcos. Reporting templates differ. The CFO at the holding level builds the quarterly board pack in a thousand-cell Excel against six different general ledgers because the portco accounting systems were never normalized after acquisition. AI is being added on top of a foundation that does not yet exist.
What this looks like across the eleven roll-ups we audited in Q1: every fund had at least one AI vendor contract signed at the holding level, with median spend of $40K to $80K annually and average utilization at the portco level under 20 percent. Three out of eleven funds had hired or were recruiting an AI Officer. None of them had given that person a portco-level reporting line. Zero funds had standardized multi-portco close cadence. Every single one was still consolidating manually on Monday mornings.
The workflows that actually got deployed and used at the portco level were not the AI Officer's pet projects. They were small, scoped, owned by an operator who picked up a tool, deployed it, and would not let it go.
The pattern is the same one we have seen with every preceding platform wave at PE firms. ERP standardization in the 2010s. Data warehouse consolidation in 2020 to 2022. Each wave came in at the holding level with a clear strategic mandate, ran into the operational reality that each portco was acquired with its own systems and processes, and ended up with a half-implemented standard that nobody used. The AI wave will follow the same arc unless the deployment model changes.
What is actually working in the portcos that have deployed AI well: small, workflow-specific tools deployed by an operator with budget authority and a problem she cares about. Not platform purchases. Not central mandates. The general counsel at one portco who deployed a contract-review GPT in two weeks. The CFO at another who put in a multi-state nonresident-partner allocation tool because she got tired of redoing it every quarter. The head of operations at a third who automated subcontractor invoice reconciliation because she was the one absorbing the manual work.
The pattern: the workflows that get pitched as 'platform AI' are usually the ones that look impressive in a board deck. The workflows that get deployed and used are usually the ones an operator picked up because she was tired of doing them manually. Holding-level AI strategy should chase the second category and stop trying to standardize the first.
What we are watching for in H2: how many funds reorganize their AI Officer role to report to portco operators instead of group strategy. The ones that do will see real adoption. The ones that do not will see another half-implemented platform purchase in 2027.
