The complex challenges mid-market businesses must address before AI can generate sustainable returns

AI has stopped being a conversation about the future. Gartner is projecting global AI spending to hit roughly $2.5 trillion by 2026, and Stanford AI Index confirms what anyone working inside mid-market businesses already senses: the technology has moved out of innovation labs and into core operations.

The momentum is real. What’s less clear is whether it’s actually working.

Plenty of organizations have run successful pilots. Far fewer have changed how they fundamentally operate. That gap shows up most painfully in mid-market and private equity environments, where the scorecard isn’t “did we launch something interesting”, it’s margins, growth, and capital efficiency.

The stack you built to survive is now working against you

Here’s the thing most AI vendors won’t tell you: the average mid-market company was never built for this. Its technology stack is a layered history of good decisions made under pressure,  an ERP to handle transactions, a CRM to get visibility, custom tools built when the team was scaling fast and needed something that worked now. None of those decisions were wrong. They just weren’t made with AI interoperability in mind, because nobody was thinking about that yet.

What happens over time is that complexity quietly accumulates. Data sits in separate systems. One department’s performance metrics don’t match another’s. Even something as basic as a customer identifier might not be consistent across platforms. Legacy tools hold years of institutional knowledge but were never designed to talk to each other. When you introduce a sophisticated AI model into that environment, you’re not solving those problems, you’re running advanced analytics on top of a foundation that was never built to support them.

This is where most initiatives stall. I’ve seen it repeatedly working with mid-market companies: AI projects disappoint not because the technology is weak, but because the organization underneath it is fragmented. AI doesn’t fix a broken operating model. It exposes one. And if you’re willing to look honestly at what it’s revealing, that’s actually an opportunity, but only if you respond with real redesign rather than another layer of workarounds.

The IMF and others have made a similar point at the macro level: AI-driven productivity gains depend less on having access to the technology and more on whether organizations are structurally ready to absorb it. The same logic holds company by company.

Process design is where this gets decided. AI reflects the quality of the decisions systems around it, it amplifies what’s already there, good and bad. Disciplined workflows with clear ownership get faster and more consistent. Diffuse accountability and metrics that don’t connect to financial outcomes just get scaled. The dysfunction doesn’t disappear; it compounds.

If it can’t move the numbers, it doesn’t matter

In PE-backed companies, there’s no patience for ambiguity here. AI isn’t a technology experiment, it’s a value creation tool. If it can’t move EBITDA, accelerate revenue, or cut structural costs within a defined window, it’s a distraction. The questions leadership should be asking before any broad deployment are blunt ones: What financial driver are we targeting? What does success actually look like, in numbers? When should we see it? That framing turns AI from an innovation story into a performance program, which is the only version that matters.

Integration is the next pressure point. Predictive insights only create value when they feed into live decisions. A pricing engine needs to connect to billing. Customer risk signals need to update CRM workflows on their own. Operational forecasts need to actually influence procurement and scheduling, in real time. Most mid-market legacy systems weren’t built for that level of connectivity. So embedding AI isn’t really a deployment problem, it’s a systems architecture problem. You’re not adding a layer on top of existing processes. You’re rethinking how the whole thing fits together.

The people problem nobody budgets for

Organizational alignment is the piece that gets underestimated most. When AI changes how decisions get made, and it does, it also changes who makes them, how information moves, and where accountability sits. If incentive structures and cross-functional communication don’t evolve alongside the technology, adoption stalls and teams retreat into their own corners. But when alignment actually happens, the payoff is real: faster decisions, sharper capital allocation, growth that doesn’t require proportional headcount increases.

These challenges, data coherence, integration, financial alignment, decision discipline, organizational buy-in, don’t respond well to being tackled one at a time. The companies that are getting this right are building a coherent operating model, not a checklist of AI initiatives.

Access to AI tools will keep getting cheaper and easier. That’s not going to be the differentiator much longer. What will separate the companies that genuinely benefit from the ones that don’t is internal capacity, the ability to turn intelligence into consistent execution.

For mid-market leaders, the real opportunity isn’t just deploying AI. It’s building a company where AI is woven into how work actually gets done, into workflows, metrics, governance, and culture. That’s a harder thing to copy than any particular tool.

The companies that figure that out won’t just be faster adopters. They’ll have built something that compounds over time.

Cesar DOnofrio is CEO and co-founder of Making Sense, where he leads digital modernization initiatives for mid-market and private equity-backed companies. Over the past two decades, he has worked at the intersection of software engineering, operational strategy, and AI-driven transformation.