Hg now employs over 80 AI engineers, product managers, and designers who embed directly in portfolio companies to ship agentic products.¹ Apax has built 370 custom GPTs for everything from analyzing CIMs to drafting IC memos.² Vista has an "internal army" deploying AI across 85-plus portfolio companies.³ Carlyle's CIO predicts that within five years, "deal teams think like technologists because they are powered by AI agents."²

This raises a question for every mid-market GP: if the largest firms in the industry are reorganizing themselves around AI at the fund level, what is your plan to keep up?

The Org Chart Problem

The standard answer is "we'll hire a head of AI": a technical lead who can go figure it out. But Blackstone's Rodney Zemmel, who leads their operating team, argues that misses the point entirely: "There is a massive difference between business teams and tech teams working together and co-owning and developing AI solutions, versus the business trying to set the goals and requirements and tech going off to deliver on them. That old nineties style IT approach doesn't work with AI."⁴

Inflexion's Alex Mathers puts it more concretely: "It isn't just a case of hiring an AI lead. It's a team game. You need a combination of skills ranging from project management, to change management, as well as newer skills around how you work with the tech itself, whether that's how to build agents using LLMs or more classic data science."⁴

The problem with "hire an AI person" isn't the hire, it's the assumption that AI can live in a box on the org chart. When Hg embeds Catalyst engineers alongside portfolio company management teams to ship agentic products in eight weeks, that work is core to the investment thesis. It touches diligence, value creation planning, product strategy, and exit positioning simultaneously. You can't silo those functions under a CTO and expect results.

What Mid-Market Firms Should Build Instead

You're not going to replicate an 80-person incubator, and you shouldn't try. The solution is putting AI capability where the investment decisions happen, not in a silo beside them. A few structural changes create disproportionate leverage at mid-market scale, and most cost less than a single senior hire.

One AI-fluent operating partner, not a data science team. The industry is moving away from "hiring expensive, hard-to-retain data scientists toward full-stack AI engineers and curated consultant ecosystems that can deliver results more efficiently."⁵ For a mid-market fund, the highest-leverage hire is someone who can sit in a deal review, assess a target's AI maturity as part of diligence, and then scope a realistic 12-month implementation plan for the portfolio company post-close. That's an operating partner with engineering fluency — not a chief data scientist who reports to nobody on the investment team.

Fractional AI specialists on retainer, deployed against portco use cases. Permira has dedicated AI experts in healthcare, services, and process industries.⁴ A $2–5B fund can mimic that approach with two or three specialist consulting relationships scoped to specific sectors or workflows. The key is that each engagement builds institutional knowledge; the patterns from one portco deployment carry to the next, which is exactly the compounding advantage the mega-funds get from their in-house teams.

AI milestones written into portco management incentive plans. Avoid "adopt AI" as a vague value creation plan line item. Set specific, measurable targets like AI-sourced lead volume, support automation rates, or engineering velocity gains, tied to the same bonus structures that already drive every other operating metric. PE is the industry that proved incentive alignment works. The firms that apply that logic to AI adoption, rather than treating it as a side initiative management runs on goodwill, will see adoption rates that look nothing like the industry average.

AI maturity as a standing diligence and IC item. What is the target's data infrastructure? How automatable are its core workflows? What does the 12-month AI roadmap look like post-close, and what will it cost? If these questions aren't part of every IC memo, you're underwriting without a complete picture of where value creation will come from over the next hold period.

The mega-funds' advantage isn't that they have more AI engineers. It's that they've made AI a structural part of how the fund operates, from sourcing through exit. A mid-market GP can build that same muscle with a fraction of the headcount, provided the commitment is organizational rather than cosmetic.