The hiring maths
Say you run value creation across a portfolio of twelve companies, each doing £20m to £150m of revenue. Your LPs are asking about AI. Your ICs are pricing AI capability into new deals. The obvious answer is to hire for it in each portco.
Across twelve companies, the cost quickly becomes impractical. A credible head of AI who can set direction and build costs £150k to £250k plus equity, if you can find one. Most portcos also lack a full-time role's worth of AI leadership work. They have three months of intense build, followed by maintenance and a steady flow of new projects.
The full-time hire is often too expensive, underused or both. Many portcos therefore end up with nobody in the role, leaving the AI line in the value creation plan theoretical.
The supply side makes it worse. The engineers who can build production AI systems mostly go to the labs and the AI-native startups, and the recruiters we work with see no blue-chip talent pool forming for AI roles inside funds or portcos. Twelve searches for people who barely exist is not a plan.
What portcos actually need
The actual need comes down to four things:
- Prioritisation: someone who can look at the business and rank AI opportunities by P&L impact rather than novelty.
- Build capacity: people who ship production systems, not proofs of concept.
- Adoption: getting the operating team to actually use what gets built.
- Continuity: someone accountable across quarters, so the capability compounds instead of resetting with each project.
Only build capacity requires a heavy time commitment, and that demand is uneven. The rest is judgement applied regularly, but not continuously, which makes a shared resource a better fit than a full-time hire.
The shared model
The model that works across a portfolio looks like this:
One partner, many portcos. A single external team serves the portfolio, carrying context between companies. The pricing pattern that worked in one portco gets adapted for the next in weeks, not discovered from scratch. The fund pays for outcomes across the portfolio rather than salaries inside each company.
Fund-level prioritisation, portco-level delivery. The fund sets the standard: every portco gets an opportunity scan, and builds are ranked by expected P&L impact across the whole portfolio. Capital flows to the best opportunities regardless of which company they sit in. Delivery then happens inside the portco, with the portco's team, on the portco's systems.
Fractional leadership where it's needed. Some portcos, usually the larger or more data-rich ones, justify a fractional head of AI: a day or two a week of senior direction, backed by the shared build team. We wrote a fuller breakdown of when that role makes sense in our piece on the fractional head of AI.
Some funds are already running this structure. One mid-market US firm of about fifteen people has hired a single head of AI for the entire portfolio: he works with each management team before and after close to pick the top three to five AI focus areas, and external builders handle everything after scoping, from engineering through to adoption. One salary buys portfolio-wide prioritisation, and build capacity is bought only when a project is live.
Why funds are structurally suited to this
This model is hard for a standalone mid-market company to buy and easy for a fund to orchestrate, which makes it a genuine source of alpha rather than something every business can replicate.
The fund aggregates demand that no single portco could justify and applies consistent delivery discipline. It also retains the pattern library: after the third portco, its AI partner has seen enough of the portfolio's operations for each new build to start from precedent.
We built our practice around this structure. We were investors ourselves before this, at firms including 3i, Apax and Deutsche Bank, so the prioritisation conversation happens in fund language: hold period, value bridge and exit story. Delivery happens inside the portco, in production code.
The question to ask yourself
If your portfolio AI plan currently depends on each portco hiring its own capability, ask when that last worked for any specialist function. Portcos don't each hire a head of pricing or an ESG director either. The fund provides or brokers that capability because the economics only work at portfolio level.
AI has similar portfolio economics. A shared capability starts accumulating operating knowledge each quarter, while a plan based on twelve separate hires cannot begin until every company can recruit and support the role.