By Cesar D’Onofrio, Co-founder and CEO of Making Sense
Over the past two decades, working alongside mid-market executives, I’ve heard the same frustration again and again: ‘We invested in AI, but nothing really changed.’ The numbers back this up.
According to Gartner, global investment in AI is expected to reach approximately $2.5 trillion by 2026. Despite this momentum, many mid-market and private equity-backed companies are struggling to generate meaningful returns from their AI investments; most remain stuck at the experimentation stage, with no measurable impact.
The challenge is that achieving ROI requires much more than deploying AI. It demands an operational environment capable of integrating and acting on the intelligence these systems produce. The reality is that the technology works, but the organization surrounding it often isn’t prepared to support it.
Mid-market companies face a different reality
Over the years, companies built technology ecosystems to solve immediate business needs, for example, an ERP system to manage transactions, a CRM to provide visibility into customer relationships, or even custom applications were developed for evolving processes. Each decision was logical and often necessary at the time.
However, these systems were rarely designed to operate as an interconnected environment capable of supporting modern AI initiatives. As businesses grow, complexity accumulates: data becomes dispersed across multiple platforms, information standards diverge, and departments develop different definitions of success. At the same time, critical institutional knowledge remains trapped within legacy applications that were never intended to communicate with one another, creating a fragmented foundation that makes AI adoption significantly more difficult.
So in these common scenarios, AI exposes the complexity that already exists. When workflows are fragmented, ownership is unclear, and teams operate under competing priorities, advanced models will not compensate; they will tend to amplify those inefficiencies, leading AI initiatives to lose momentum. Working alongside mid-market organizations, I have identified that the underlying operating model lacks the cohesion needed for AI to deliver meaningful results, and the productivity gains associated with AI depend not only on access to the technology but on whether organizations possess the structural readiness to integrate it effectively.
The missing link between AI and financial outcomes
The pressure to leverage AI is coming from every direction: boards, investors, and competitors, so organizations tend to jump right into launching a pilot before anyone stops to ask whether the company is actually ready to sustain AI-driven transformation.
The real gap between using the technology and being ready for it is wide, and in the mid-market, where internal tech teams are smaller and budgets are tighter, it’s even wider.
More often than not, these companies inherit outdated ERP systems, disconnected databases, and none of the seamless integration that cloud-native startups or enterprise firms benefit from.
For example, Vetsource, a leading home delivery partner for vets and pet owners, faced this reality when rapid post-COVID growth exposed the limits of its legacy systems. It had a short window to modernize or risk losing market share. To support scale, the company deployed a comprehensive cloud-based ecosystem that included upgraded e-commerce and payments infrastructure. This replaced manual processes with automated, AI-driven workflows. The result was stronger operational efficiency, helping Vetsource double its customer base of paying customers and achieve a 900% increase in profitability.
Vetsource’s software modernization story is highly relatable for any company, as software has been one of the top asset classes in PE-backed portfolio companies for the last decade. SaaS businesses offered recurring revenue, high margins, sticky customers, and clean growth metrics. Now, AI is forcing a reassessment of that thesis, and fast.
So, before scaling AI investments, leadership teams should be able to answer straightforward questions, at the company level, “How will value creation be measured?” and “What timeline is realistic for achieving results?” and for investors, “Which financial metrics will improve?” and of course, “How does AI support EBITDA growth, revenue expansion, or structural cost reduction?”
The honest answers to those questions turn AI from an innovation story or another technology deployment into a strategic performance program.
Funding cycles deepen infrastructure gaps
Private equity traditionally operates on a 3-to-5-year investment horizon, and portfolio performance is measured by EBITDA improvement—not long-term infrastructure investment. Meanwhile, most organizations take 2 to 4 years to achieve satisfactory ROI from AI initiatives. For a PE-backed company already two years into a fund cycle, ROI may only materialize after the company is sold.
How do you underwrite a five-year deal if you can’t predict what happens to the software in the next one or two? Even those who understand AI cannot always see far enough ahead to bet confidently on it, so they pull back.
Investors who entered tech for strong returns are shifting back toward more predictable business models, such as transportation, manufacturing, healthcare, and legal services. But these are the very businesses most likely to benefit from AI. So investors who view AI not as standalone products but as systems embedded into key operations like pricing, scheduling, and service delivery will find advances in these traditional models over those purely focused on AI-centered solutions.
The real challenge lies in making AI work in traditional industries. Most executives feel confident in why they pursued AI: to stay competitive, improve margins, or accelerate innovation. But when it comes to answering “how much did this really move the needle?”, things get less straightforward.
The foundations before the pilot
Getting AI-ready is less about selecting the right tools and more about laying the groundwork that allows any integrated platform or solution to improve processes from day one, so those that get it right tend to have four foundations in place:
- Centralized data infrastructure: They unify disparate data sources into a single source of truth that enables reliable decision-making and consistent reporting. In practice, this means implementing a data integration layer or warehouse (e.g., Snowflake, BigQuery) that pulls data from CRMs, ERPs, and spreadsheets and consolidates it into a single, structured database.
- Legacy system modernization: In many mid-market firms, core operations still run on older ERP and finance systems that were never built for real-time data exchange or AI integration. This creates complex manual processes that often keep businesses stuck in a reactive mode. A modernization plan aligns legacy systems with modern cloud and data architectures to enable automation and scalable integration.
- Workflow alignment: This involves redesigning business processes around human-AI collaboration, ensuring clear ownership and defining handoff points before deployment. Establishing objectives and ownership upfront enables AI to be embedded into workflows, accelerating execution and ultimately creating measurable business value.
- Speed to value: Within AI implementation workflows, the biggest mistake I see is over-scoping. Companies often build comprehensive AI strategies that take months to plan and even longer to execute. But by the time they launch, the landscape has already shifted. The goal should be to demonstrate measurable impact in weeks, not months. Early wins build momentum for later stages.
- Cross-functional teams: Successful organizations typically establish teams that bring together operations, data, and technology to oversee this effort through a structured framework. A U.S Chamber of Commerce workforce report found that 61% of mid-market firms facing staffing challenges plan to invest in AI tools. But skills training remains the missing piece needed to make those investments productive. This is precisely why we’ve built an AI Enablement and Learning Program to address that gap.
Sustainable AI value requires a holistic approach
The most common implementation mistake is treating AI adoption as a series of independent projects. From our experience working with companies at different stages of AI adoption, achieving meaningful outcomes is not about building collections of AI experiments. They are developing operating models designed to support AI at scale.
For mid-market leaders, the opportunity is bigger than deploying a new tool. It is about creating an environment where AI becomes embedded within workflows, performance metrics, governance structures, and company culture. That kind of transformation is far more difficult to replicate than any individual technology platform.
The companies that succeed will not simply adopt AI faster than their competitors. They will build organizations capable of compounding AI value over time. Ultimately, the future belongs to businesses that view AI not as a technology initiative, but as an opportunity to rethink how knowledge, decisions, and people work together.
Cesar DOnofrio is the 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.
Featured image: Growtika via Unsplash+
Disclosure: This article mentions a client of an Espacio portfolio company.
