By Chris Culbert, Principal, JMAN Group
FinTech moves fast. News is everywhere, clarity isn’t.
FinTech Weekly delivers the key stories and events in one place.
Click Here to Subscribe to FinTech Weekly's Newsletter
Read by executives at JP Morgan, Coinbase, BlackRock, Klarna and more.
Private equity has always been a business of judgement. Capital structure amplifies returns, but interpretation determines them: which pricing lever to pull, which cost base to reshape, which segment to prioritise. For decades, those decisions were formed through experience, debate, and periodic review of aggregated financial performance.
That model worked in a forgiving environment. It works less comfortably now. Higher interest rates, slower deal velocity, and tighter valuations reduce the margin for interpretive error. Multiple expansion no longer compensates for operational leakage. Precision inside the portfolio matters more than financial engineering alone.
Artificial intelligence is often framed as an analytics accelerator. The adoption numbers support that narrative. Assets managed through algorithm-driven and AI-enabled platforms are projected to approach $6 trillion in the coming years, and a majority of private equity firms report active investment in AI across portfolio oversight and data infrastructure.
Yet the way AI is entering portfolio companies is not through sweeping technology overhauls. It is entering more quietly, through the embedding of small, technically sharp data science teams directly into portfolio operations. I refer to these teams as “AI hyenas.”
The term is deliberate. Hyenas are adaptive; they operate close to the ground and survive by detecting variance others overlook. These embedded teams behave similarly. They work at transactional depth rather than relying on summarised reporting. Their advantage is not speed alone but resolution. They surface dispersion in pricing, cost structure, demand patterns, and working capital dynamics that traditional operating reviews struggle to detect at scale.
At first glance, this appears to be tactical optimisation layered onto the existing operating landscape
Consider pricing. Traditional reviews rely on segment averages and periodic executive debate. Embedded AI teams build models at granular levels, identifying micro-segments where pricing power exists or where margin erosion is occurring relative to demand conditions. What once required extended analysis now arrives as a quantified signal with defined confidence ranges.
The same logic applies to demand forecasting and capital efficiency. Machine learning models integrate internal performance data with external signals, simulate scenarios and refine projections dynamically. Inventory adjusts with greater accuracy cash conversion tightens and variance that previously dissipated unnoticed becomes visible.
This is the visible layer of change: operational analytics becomes sharper, response becomes faster and incremental value is extracted more consistently.
The more consequential shift, however, is less obvious.
As model-generated recommendations become embedded within pricing discussions, forecasting cycles, and capital allocation reviews, they begin to alter how the operating landscape functions. Decisions are surfaced differently, signals enter earlier and response cycles compress. The architecture of decision-making begins to evolve.
Historically, management teams discovered patterns through discussion and interpretation; insight preceded action. Increasingly, quantified recommendations enter the process before collective debate. The question shifts from “what is happening?” to “how should we respond to this signal?”
That shift is not about automation. It is about agency.
Authority inside the operating landscape begins to redistribute. Leaders move from discovering patterns to defining thresholds, escalation points and override conditions. Judgement does not disappear; it changes position.
This is where governance moves from overhead to operating design.
In an AI-enabled portfolio company, governance determines how decision rights are allocated between human judgement and system-generated recommendation. It defines who owns a signal, how it is validated, when it can be overridden, and how outcomes feed back into future models. Without that clarity, embedded analytics remain peripheral. With it, they become structural.
Many firms have historically attempted to codify operational best practice into playbooks. In stable environments, that approach can scale consistency. In environments where signal shifts rapidly, static playbooks struggle. AI-enabled operating models do not eliminate discipline; they require a different kind of discipline built around adaptive thresholds, governed decision rights, and continuous feedback rather than fixed procedural templates.
Sponsors who rely solely on codified operating playbooks may find themselves optimising for a landscape that is already receding. Those who design operating models around live signal and deliberate agency allocation will adapt faster.
Research across financial services consistently identifies governance and integration (not model accuracy) as the primary barrier to scaling AI. The constraint is rarely technical; it is organisational. It is ambiguity about how AI sits inside the operating landscape.
AI hyenas succeed because they are adaptive. They embed within existing workflows rather than attempting wholesale redesign, generating signal where it matters most. Sponsors who extract durable advantage recognise that operational analytics is only the visible layer. The deeper evolution occurs when governance deliberately reshapes the operating model around that signal.
This evolution has direct implications at exit.
Buyers increasingly interrogate not only performance outcomes but the robustness of the operating landscape that produced them. Granular and auditable operational data demonstrates that pricing discipline, demand forecasting, and capital efficiency are governed capabilities rather than episodic improvements.
A mature data environment reduces diligence friction. More importantly, it signals resilience, showing that performance is not dependent on individual judgement alone, but on structured decision architecture capable of sustaining performance under new ownership.
Financial engineering will remain part of private equity. The next frontier of value creation lies in how signal flows through the organisation, how authority is structured in response to that signal and how governance transforms from compliance to agency management.
The AI hyena is the adaptive mechanism through which that transition begins. They enter the existing operating landscape quietly, extracting value at transactional depth. Over time, it reshapes how decisions are formed, governed, and defended.
The firms that recognise both layers - the immediate operational gains and the underlying redistribution of agency - will not simply optimise margins; they will evolve deliberately.
In a market where precision compounds, that evolution becomes decisive.
