AI adoption in middle market companies involves more than adding new tools. For private market investors, the question is whether a company can improve defined workflows while controlling cost, data and execution risks. This article examines the operating changes managers can assess and the evidence needed to distinguish experimentation from measurable results.
A note on perspective: this article presents opinions and a forward-looking thesis. It is not a forecast, and the outcomes it describes may not occur.
Could Companies That Use AI Capture Lasting Value?
In Crystal's view, the businesses that use AI well, not only those building it, may capture a meaningful share of its value. It is a conditional perspective, not an established finding.
As we read some earlier technology shifts, such as the railroads, many durable beneficiaries were the businesses that reorganized around the new technology rather than the companies that built it. Whether that pattern repeats with AI is uncertain. If AI creates value in mid-sized businesses, the open question is who will capture it: the companies and managers who redesign operations around the technology, or those who treat it as a line item.
Are you a financial advisor? See how the Crystal platform works.
How Does AI Adoption Differ From Measurable Operating Results?
Adoption means a company uses AI tools; operating results mean a defined workflow measurably improves, net of the cost of getting there. The first doesn't guarantee the second.
A company can roll out AI tools widely without changing how work gets done. Results are more likely to show up, in our view, when a company redesigns a specific workflow around what the technology does well, sets a baseline, and measures the change in time, cost, quality or revenue. For investors, that distinction, between using AI as a tool and rebuilding a process around it, is the starting point for evaluating any claim about AI.
What Can Help or Hinder AI Adoption in Middle Market Companies?
Simpler systems and faster decision-making can help some middle-market companies adopt AI-enabled workflows; thinner teams, limited budgets and data quality can hold them back.
Definitions of the middle market vary, most often by annual revenue, so the term covers a wide range of businesses. Some of them carry less complex legacy software than large enterprises and can make decisions with fewer layers of approval, which may let them change workflows more directly. Others face the opposite: scattered or poor-quality data, small technology teams, limited budgets for implementation, and dependence on a few key people.
- Can help: simpler systems, short decision chains, owners or sponsors willing to redesign processes, clear operational metrics.
- Can hinder: poor data quality, thin technology and data skills, limited budgets, privacy and security obligations, and change fatigue.
What Does an AI-Enabled Operating Model Look Like?
An AI-enabled company redesigns workflows and decisions around what the technology makes cheap, rather than bolting AI onto existing processes.
| Tools added | AI-enabled | |
|---|---|---|
| Approach | Adds AI tools to existing processes | Rebuilds workflows around AI capabilities |
| Data | Unstructured data stays in documents and inboxes | Contracts, emails and records become searchable, decision-ready data |
| Decisions | Same decision points, slightly faster | Pricing, forecasting and customer analysis that were previously too costly become routine |
| Governance | AI used mainly for productivity | AI also used to challenge assumptions and stress-test decisions |
The practical test is at the level of individual workflows. The examples below are illustrative, not results from any company:
| Illustrative workflow | Intended benefit | Evidence a manager should provide | Relevant risks |
|---|---|---|---|
| Reviewing contracts | Faster review and fewer missed terms | Review time and error rates before and after, with human sign-off retained | Errors in outputs, confidentiality, over-reliance without review |
| Organizing customer records | Usable data for pricing, service and retention analysis | Data-quality measures and how the cleaned data is actually used | Privacy and security obligations, poor source data |
| Supporting demand forecasts | Better inventory and staffing decisions | Forecast accuracy against the prior method over several periods | Model drift, unusual market conditions, ongoing running costs |
Across all of these, data quality, privacy, security, human oversight and ongoing operating costs determine whether a pilot becomes a durable change. In our view, the lasting advantage of AI lies in improving human decision-making, not removing people from it.
What Does AI Adoption Cost?
The cost of AI adoption goes beyond software licenses: integration, data preparation, process change, training and ongoing running costs can all be significant.
Putting AI to work typically involves connecting it to existing systems, cleaning and organizing data, redesigning processes, training staff, and paying for usage, monitoring and security over time. Those costs can offset part or all of the expected benefit, and they often continue after a pilot ends. For an investor, a credible AI plan sets out the full cost alongside the expected benefit and shows how both will be measured.
What Are the Risks of AI Adoption in Middle Market Companies?
AI initiatives can stall at the pilot stage, cost more than expected, create data and security exposure, or disrupt a company's own business model.
- Execution. Pilots can stall without producing measurable results, and smaller companies may lack the talent to carry projects through.
- Data quality. Outputs are only as reliable as the data behind them.
- Privacy and security. Customer and employee data, and the tools that process it, create new obligations and exposures.
- Oversight and governance. Errors can go unnoticed without human review, and governance for more autonomous systems is still developing.
- Ongoing costs. Usage, monitoring and maintenance costs can erode the benefit.
- Disruption. AI could weaken some middle-market business models rather than strengthen them.
- Valuation. Prices for businesses seen as AI beneficiaries can get ahead of fundamentals.
How Should Private Market Investors Evaluate AI Adoption?
By asking for evidence at the workflow level: a baseline, a measured change, the full cost, and who oversees the result.
Asking a manager "are you using AI?" isn't very informative. More useful is evidence that distinguishes experimentation from results: which workflows changed, how the before-and-after was measured, what the change cost to implement and run, and how much of any improvement can reasonably be attributed to AI rather than other changes. In Crystal's view, a manager's operating capability, not access to AI as a theme, is the more important part of evaluating middle-market private equity and credit, and that view could prove wrong.
Questions to Ask a Middle-Market Manager About AI
- Which workflows have your portfolio companies redesigned around AI, rather than simply adding tools?
- Where can you show a measurable operating benefit from AI in a portfolio company's results today?
- How do you assess AI as a risk, including disruption to portfolio companies' own business models?
- What operating resources do you provide to help companies implement AI, and how do you build that capability across funds?
- How do you use AI in your own investment process, including to challenge your assumptions?
For more on how middle-market managers create value through operations and add-on acquisitions, see our article on lower middle market private equity.
Conclusion
AI adoption in middle market companies is easy to claim and harder to demonstrate. Simpler systems may help some companies redesign workflows, while data quality, thin teams and ongoing costs can hold others back. For private market investors, the useful questions concern specific workflows, measured results and full costs, with the caution that any thesis about a fast-moving technology may not play out as expected.
Frequently Asked Questions
What is AI adoption in middle market companies?
It is the process by which mid-sized businesses put AI to work in their operations. Adoption alone means using AI tools; the more meaningful step is redesigning specific workflows and measuring whether they improve, net of cost.
How can investors assess whether AI is improving operations?
By asking for workflow-level evidence: a baseline, the measured change in time, cost, quality or revenue, the full cost to implement and run the change, and who oversees the results.
What costs and risks can offset AI-related benefits?
Integration, data preparation, training and ongoing usage costs can offset expected benefits. Risks include stalled pilots, poor data quality, privacy and security exposure, weak oversight, and disruption to a company's own business model.
What does an AI-enabled operating model look like?
An AI-enabled company redesigns its workflows and decisions around what the technology makes cheap, such as turning unstructured data into usable information, rather than simply adding AI tools to existing processes.
What should investors ask private equity managers about AI?
Useful questions include which workflows portfolio companies have redesigned, where AI has produced a measurable operating benefit, how the manager assesses AI-related risks, what operating resources it provides, and how it uses AI in its own investment process.
About the examples: This article presents opinions and forward-looking views for educational purposes. It contains no performance figures and is not a prediction or a recommendation to invest in any company, sector, fund or Crystal Portfolio.