The differentiator is not whether a business uses AI. It is knowing where it belongs.
There is no shortage of AI ambition in the middle market right now. Nearly every company has a pilot underway, every software vendor has a new feature to demo, and every board has a slide that promises transformation. Yet despite all the activity, one question remains surprisingly difficult to answer:
Where does AI actually create competitive advantage?
The cost of getting that decision wrong is now well documented. A widely cited 2025 study out of MIT found that roughly 95% of enterprise generative-AI pilots delivered no measurable impact on the bottom line, despite tens of billions of dollars invested. The technology itself was rarely the problem. The technology itself was rarely the problem. The report pointed instead to a gap between experimentation and systems capable of learning, adapting and integrating into how work actually gets done.
At Kingsley Gate, we spend much of our time alongside the operators who create value inside businesses, and one theme has become impossible to miss. The leaders pulling ahead are not the ones deploying the most AI. They are the ones with a clear filter for where it belongs.
To better understand that discipline, Alison Woodhead, Senior Partner at Kingsley Gate, spent time in discussion with Robert Root, Transformation Partner at Southfield Capital. As an operating executive supporting founder- and manager-owned business-services companies, Root evaluates these decisions across portfolio companies every day. His framework is remarkably practical.
For Root, the opportunity may be particularly significant in the middle market. Historically, smaller businesses have lacked the technology budgets and engineering resources of larger competitors.
AI has the potential to change that equation, giving companies that move thoughtfully an opportunity not simply to catch up, but to leapfrog larger competitors.
“It is the topic du jour. The reality of AI right now is that everything is AI. Everyone is throwing AI on everything. People need to be very strategic and thoughtful about where they use it.”
— Robert Root – Transformation Partner at Southfield Capital
His thinking ultimately comes down to one central principle: Don’t start with AI. Start with what makes your business unique.
Start with the data moat, not the software menu
The first question Root asks is not Where could we use AI? The answer to that is almost always everywhere.
Instead, he asks a much harder question: What does our business know that no one else does?
That is what he calls the data moat. A data moat is the proprietary information a company generates simply by operating its business. Anyone can buy software and AI models are becoming increasingly accessible. The information in a company’s data moat is unique. It reflects years of customer interactions, operational history, workflows, outcomes, and institutional knowledge that competitors cannot easily replicate. When AI is applied to that proprietary data, it becomes much harder to copy. AI turns that proprietary information into differentiated products, services, and insights.
The distinction is easier to see through two very different examples.
Take invoice processing. Nearly every ERP platform now offers AI capabilities to read invoices, extract data, and automate approvals. For most companies, buying that capability is exactly the right decision. Paying invoices is not a competitive advantage. There is no proprietary insight being created, and therefore little reason to build anything custom.
Now consider a fleet-management company responsible for maintaining industrial forklifts. Every maintenance visit generates information unique to that business: equipment hours, repair histories, operator behavior, component failures, replacement cycles, contract compliance, and lifetime ownership costs. Over time, that becomes proprietary operational intelligence. That is the moat.
Applied to that data, AI does far more than automate paperwork. Customers pay for the lowest total cost of ownership and the highest uptime across their fleet, and the maintenance stream is exactly the raw material required to deliver both. It can audit invoices against service agreements, identify contract violations automatically, predict equipment failures before they occur, recommend replacement timing, and ultimately become an advisory platform that helps customers reduce downtime and lower total cost of ownership. None of that comes from the AI model itself. It comes from the data only that business possesses.
As Root explains: “Going back to this data moat is really thinking about what your business is providing and what is unique about it. Start there, not from the feature list of the software you already own.”
That distinction changes the conversation entirely. The competitive advantage isn’t AI. Increasingly, AI is becoming a commodity. The competitive advantage is proprietary data.
It is a discipline any leader can apply, in any sector. The moat is rarely the flashiest part of the business. But it is the part no competitor can replicate.
The efficiency trap, or when “faster” makes things slower
The most common mistake is the mirror image of that discipline: treating every AI-enabled efficiency as progress. Companies add AI assistants to the tools they already use, but rarely stop to ask whether they are simplifying the work or simply adding another layer to it.
That distinction matters. A Harvard Business Review study found that employees toggle between applications roughly 1,200 times a day, losing close to four hours a week simply reorienting after each switch. In middle-market businesses, where individuals often wear multiple hats and work across multiple systems, adding another layer of disconnected tools can compound the problem.
AI should reduce that fragmentation, not add to it. The goal is not to make five separate steps marginally faster. It is to ask whether those five steps need to exist separately at all. Otherwise, the tool designed to save time can end up consuming more of it.
AI is not a fourth lever. It connects the other three.
Operators have long framed value creation around three levers: people, process and technology. It is tempting to think of AI as a fourth lever. Root sees it differently. AI’s real potential is in connecting the other three, allowing businesses to rethink how work gets done across functions, systems and processes rather than simply making individual tasks faster.
Applied narrowly, AI can make an existing task faster. Applied more strategically, it can connect people, processes and systems that have historically operated separately, allowing leaders to rethink how work gets done from end to end. The opportunity is not simply to optimize individual steps, but to ask whether those steps, handoffs and functional boundaries need to exist in the same way at all. That kind of redesign requires understanding how work actually happens, not how it appears on an org chart or process map.
That does not mean reinventing everything. For work that is largely transactional or does not contribute to what makes the business distinctive, the AI already embedded in existing software may be more than enough. A CRM that helps salespeople organize their pipeline and follow up more effectively does not require a custom solution.
The calculus changes as the work gets closer to the core of what makes the business valuable. There, AI can connect proprietary data with the people making decisions and the processes through which value is delivered. Used this way, the goal is not simply to automate work or reduce headcount. It is, as Root puts it, to “10x” the capability of people, giving them better information, better tools and the ability to make better decisions.
Sometimes the smartest AI investment is the one you don't make
Some of the hardest AI decisions are the ones where the technology clearly works, but the investment still does not make sense. Consider an AI tool that could make an HR function dramatically more efficient. The tool does everything it promises. But the HR team consists of two people, and the business will still need both of them after implementation. The efficiency is real. The economic value is not.
It is a simple example, but it gets at a distinction Root believes is often lost in the rush to adopt AI: efficiency is an outcome, not a strategy.
“I do not start with the lens of driving efficiency. That is boring to me. Cost savings is not a strategy. It is a means to an end. So, get really clear on what you want your endpoint to be.”
— Robert Root – Transformation Partner at Southfield Capital
That endpoint matters. The question is not simply whether AI can reduce time, cost or headcount. It is what the business intends to do with the capacity it creates. The most valuable applications may generate efficiencies along the way, but those efficiencies can then fund investments that strengthen the offering, improve the customer experience or create new sources of growth.
The objective is not simply to make the business cheaper to run. It is to create capacity for what Root describes elsewhere as the second- and third-order benefits of AI: new services, new business models and new market opportunities.
Do two things at lightning speed, not eight at a crawl
Deciding where AI belongs is only half the discipline. The other half is sequencing. Enthusiasm can quickly become the enemy of execution. One promising use case turns into five, then ten, with resources spread across all of them and little actually making it into the business.
Root advocates the opposite approach: choose one or two high-value use cases, move quickly, and make the results visible across the organization.
Success creates momentum. Once people see what is possible, they begin to identify opportunities in their own work. The next ideas come from the people closest to the problems, informed by what the organization has already learned.
The broader market experience reinforces Root’s point. Bain found that while most private equity portfolio companies were experimenting with generative AI, only about 20% had moved use cases into production and were seeing concrete results. The firms making the most progress were not trying to solve everything at once. They were building capabilities, focusing AI on strategic priorities and managing decisively through the uncertainty.
That kind of focus requires ownership from the top. Because AI can connect people, processes and technology across functions, no single department can own the opportunity in isolation. It needs to be a leadership priority, with the CEO setting the direction and forcing choices about where to focus.
The differentiator is rarely access to the technology. It is the leadership discipline to choose where it matters, move quickly, learn, and build from there.
The prize is a more valuable company, not a cheaper back office
Used with this kind of discipline, AI should not make businesses look more alike. It should help them lean harder into what already makes them distinctive. That brings the argument back to where Root started: the data moat.
As AI models and software become increasingly accessible, the advantage shifts to what a business uniquely knows and how effectively it can turn that knowledge into better decisions, better customer experiences and new sources of value.
McKinsey’s analysis of PE-backed companies found that those at its highest level of AI maturity traded at a median revenue multiple of 31x, while median revenue per employee was 52% higher than companies at the next level of AI maturity. The opportunity is not simply to take cost out of the business. It is to build a more differentiated and ultimately more valuable one.
That requires a different starting question. Not Where can we use AI? or even Where can AI save us money?
But: What does our business know that others do not, and how can AI help us turn that advantage into greater value?
That is the discipline that separates AI adoption from AI strategy.
This article is part of a Kingsley Gate Insights series drawing on a conversation between Alison Woodhead and Robert Root, and on Root’s April 2026 appearance on Nasdaq Trade Talks. Quotes are used with permission.
Referencias
MIT NANDA research on enterprise generative-AI adoption (“The GenAI Divide: State of AI in Business 2025”), as reported by Fortuna, August 2025.
Harvard Business Review, “How Much Time and Energy Do We Waste Toggling Between Applications?”, August 2022.
McKinsey & Company, “Beyond productivity: How AI creates value in private equity.”
Southfield Capital / Nasdaq Trade Talks, “Integrating AI into Private Equity and Investment Banking Operations,” March 2026 appearance.
English 
Español de México