.png)
Roughly one in five private equity portfolio companies have operationalized generative AI use cases and are seeing concrete results this far into AI's push across the industry. That number is worth sitting with. It means the large majority of portfolio companies are still somewhere between curiosity and pilot, even as the firms that own them are under increasing pressure to show technology as a lever for value creation, not just a line item on a hundred-day plan.
Most firms treat this as a portfolio company problem. Each company runs its own experiment, on its own timeline, with its own budget, largely disconnected from what any other company in the portfolio is doing. That approach misses the more interesting opportunity, and it is worth understanding why.
Forte Group recently worked with Xceptor, a data automation platform serving financial institutions, to move AI from ad hoc tool adoption into a structured, embedded part of how their engineering team builds software. The results were substantial on their own terms. Connector delivery time dropped 83 percent. A configuration workflow that was estimated at 26 days shipped in 6, a 77 percent reduction. Rework fell from roughly 30 percent to under 10 percent. Test scripting output quadrupled per engineer. Design prototyping returned 32 days per designer, per year. Across the engagement, 85 percent of the engineering team adopted the new way of working, well above the 50 to 60 percent target Forte and Xceptor had set going in.
What happened next is the part that matters for operating partners. Through that engagement, Forte Group was introduced to Astorg, the private equity firm behind the investment. The relationship expanded from a single agent build into a firm-wide AI programme covering Astorg's broader portfolio.
That sequence is not an accident. It is what happens when a technology partner can point to one company's proven, measured operating model and replicate it, rather than starting from a blank page with each new portfolio company.
This is the point most firms miss. The value in an engagement like Xceptor's was not a piece of software. It was a repeatable process: a way of moving from individual tool adoption, to AI embedded in the workflow, to AI executing defined steps with a human approving every gate along the way. That process transfers. A configuration builder or a delivery pipeline can be rebuilt for a different codebase in a different industry. A rework rate that drops from 30 percent to under 10 percent because the process itself changed is not specific to trade reconciliation software. It is specific to disciplined process design, and process design travels.
Firms that treat AI adoption as forty separate portfolio company problems are re-buying the same discovery work forty times. Every new engagement relearns the same lesson: individual tool adoption without process change does not move the numbers. Firms that instead identify one strong candidate, fund a disciplined build with measurable outcomes, and use that as the reference case for the rest of the portfolio are building an asset that compounds. Companies that skip that first expensive round of discovery and start from a validated playbook get to results faster and cheaper the second, third, and fourth time.
The speed difference is not marginal. The costly part of any first engagement is not the technology, it is finding out which parts of a process are ready for AI and which are not, and calibrating around the ones that were not. That discovery work took Xceptor a genuine investment of time before the results in this article were possible. A second portfolio company starting from that playbook skips the discovery and moves straight to calibration for its own context. The third and fourth move faster still, because the operating partner sponsoring the programme has now seen enough versions of it to know what a stalled rollout looks like before it stalls.
There is also a diligence and exit dimension worth naming directly. Quantified, audited outcomes, cycle time, rework rate, cost per delivery, are exactly the kind of evidence that travels well into an investment committee memo or an exit narrative. A vague claim that a portfolio company uses AI does not move a multiple. A documented 83 percent reduction in delivery time, backed by the process that produced it and reproducible at the next company, does.
This also changes how technology diligence should be scoped for new deals. Instead of asking whether a target has adopted AI tools, the more useful diligence question is whether the target has, or could build, a measured operating model: a process where outcomes are tracked, where handoffs are minimized, and where AI is embedded rather than bolted on. That question predicts future value creation far better than a checklist of which tools are licensed.
None of this works as a mandate handed down from the fund level.
Not every portfolio company is ready. Xceptor had a functioning engineering discipline before AI was layered in; a company still fighting basic process problems needs those fixed first, or AI will simply make an undisciplined process faster at producing the wrong outcome. The programme also needs an internal champion, at the operating partner level or inside the portfolio company, because the shift from tool adoption to embedded process took roughly a year even at a company already motivated to make it work.
A portfolio-wide strategy also still moves one company at a time. Each business has to build its own version of the muscle. What transfers is the operating model and the lessons learned from building it, not a plug-and-play deployment that skips the calibration work entirely. Firms should expect the first engagement to look like an investment, not a pilot with a discount. The return shows up in the second, third, and fourth portfolio company that does not have to relearn what the first one already proved.
The Astorg relationship began with one portfolio company and one operating partner willing to fund a properly measured engagement. That is a lower bar than most firms assume. Most portfolios already have a candidate: a company with a functioning engineering team, a leadership group open to changing how work gets done, and a sponsor willing to measure the result honestly rather than declare victory early. That company, not a new pilot somewhere else in the portfolio, is where the next conversation should start.
Forte Group works with private equity firms and their portfolio companies to build exactly this kind of repeatable AI operating model, and to turn one portfolio company's proof point into a portfolio-wide strategy. If you’re looking to identify your own reference case, let’s talk.
83% faster delivery. Under 10% rework. See exactly how Xceptor got there.
.png)