The true leverage of artificial intelligence is human capital
In finance, we are accustomed to thinking in terms of allocated capital, expected return and opportunity cost. Every euro left idle carries an implicit cost. Every euro poorly invested produces dispersion. Every euro well invested generates a return and, when deployed with leverage, can amplify the outcome.
The same reasoning applies to artificial intelligence as used in business.
AI becomes interesting when it stops being a technology fad and is read for what it can be: an operational lever on human capital.
Repetitive work is frozen capital
In many companies, particularly SMEs, a significant portion of people's time is absorbed by necessary but low-productivity activities: data entry, document management, standard responses, reconciliations, information synthesis, report preparation, manual retrieval of already-available content.
These are activities that do not disappear from the organisation, but can change in nature. The machine can handle the repetitive element; the person remains responsible for framing the problem, checking the output, correcting errors, interpreting exceptions and making decisions.
This is where the leverage effect emerges. For the same labour cost, a company can achieve greater productive capacity by freeing up human time from low-yield tasks and redirecting it towards higher-value activities: client relationship management, sales, analysis, oversight, design, business development and service quality.
Technology does not automatically create value. It creates potential capacity. Value only materialises if that capacity is reallocated effectively.
The Italian data: the window is still open
The most recent ISTAT snapshot shows a market still in its early stages, but accelerating. In 2025, 16.4% of Italian companies with at least 10 employees use at least one artificial intelligence technology. The figure was 8.2% in 2024 and 5.0% in 2023.
The SME figure is even more significant: among companies with 10–249 employees, AI adoption rose from 7.7% in 2024 to 15.7% in 2025. Large companies, as expected, are considerably further ahead: 53.1% already use at least one AI technology.
The most important figure, however, is a different one: 83.6% of the Italian companies covered by the survey do not yet use any artificial intelligence technology. This means that the competitive window is still open.
Today, making effective use of AI can still represent a genuine advantage. In a few years, simply claiming to use it will likely no longer be sufficient. It will be a standard capability — like the cloud, ERP systems or CRM platforms. The differentiating factor will not be between companies with AI and companies without it, but between companies that will have learned to integrate it into their processes and those that will have bolted it on as an accessory.
Those who move first are not merely purchasing a tool. They accumulate experience, redesign processes, build internal routines, develop people, collect data, and learn where automation generates value and where it merely produces noise. This organisational capital cannot be replicated in a matter of weeks.
The false myth of cost savings
Many discussions about AI begin with the topic of cost savings. This is understandable, but reductive.
If a company evaluates AI solely as a means of cutting hours or headcount, it risks missing the most important part of the argument. The real return, particularly for an SME, does not necessarily lie in doing the same things with fewer people. It lies in enabling the same people to do more useful things — better, or more closely aligned with the client.
It is enabling the same people to perform activities that are more useful, more profitable, and closer to the client.
A team member who spends less time filing documents can devote more time to client relationships. An analyst who automates data collection can focus on interpretation. A sales professional who reduces time spent on reports and updates can increase the time dedicated to revenue-generating conversations.
This is the return on human capital. It is not an abstract formula. It is a very concrete managerial choice: deciding which activities should be automated, which should remain human, and which should be redesigned from the ground up.
The opportunity cost of waiting
In finance, uninvested capital carries a cost. Even when it does not appear on the income statement, it exists. It is the return that could have been achieved through a superior alternative deployment.
The same applies to AI. A company that defers adoption does not remain in a neutral position. It continues to pay qualified personnel to carry out activities that could be lightened, automated, or accelerated. It continues to consume hours on low-value tasks. It continues to delay higher-margin activities because available time is already absorbed.
The cost of waiting is insidious because it does not immediately present itself as a loss. It does not arrive in the form of an invoice. It accumulates in the sluggishness of processes, in reduced commercial capacity, in inconsistent quality, in time spent redoing work already completed, and in the difficulty of scaling.
For many SMEs, the risk is not that AI is too expensive. The risk is continuing to underestimate how much it costs not to use it.
Leverage only works when the process is well designed
Naturally, leverage amplifies in both directions.
An AI project applied to a poorly structured process does not produce efficiency: it often produces only faster confusion. If data are disorganised, if responsibilities are unclear, if no one knows who is responsible for validating the output, if no control criteria exist, automation does not solve the problem. It makes the problem harder to detect.
The first question, therefore, is not which software to purchase. The first question is: which process do we want to improve?
Operational questions
How much time does the process consume today? How repetitive is it? How significant is its impact on financial results? What is the cost of error?
Control questions
Who is responsible for validating the output? When is managerial intervention required? How do we measure the actual benefit?
Without these questions, AI becomes an experiment. With these questions, it becomes an investment.
Human work does not disappear: it shifts
One of the most common misconceptions is the belief that automation simply reduces workload. To some extent, it may. But in practice, the nature of work often changes.
The individual performs fewer repetitive tasks, but must exercise greater judgement: checking, verifying, interpreting, deciding. Operational work decreases; cognitive work increases. This transition must be carefully designed.
If everything is entrusted to the machine without oversight, the risk of undetected error increases. If, conversely, every output must be obsessively reviewed, the benefit of automation is diminished. The equilibrium point depends on the process, the cost of error, and the maturity of the people involved.
This is why training is not an optional extra. It is part of the investment.
People must understand what AI can do, where it can go wrong, how to verify a response, when to stop, which data must not be entered, which decisions cannot be delegated, and which steps must remain traceable.
In the absence of these competencies, AI can generate a false sense of efficiency: high output volumes, high speed, low accountability.
The AI Act makes training an organisational matter
The European framework points in the same direction. Article 4 of the European Artificial Intelligence Act requires providers and professional users to ensure an adequate level of AI literacy, taking into account technical knowledge, experience, education, training, and the context of use.
For businesses — particularly SMEs — this is a significant step. Training in artificial intelligence should not be viewed merely as a compliance exercise. It is a form of productive capital. It serves to reduce errors, improve decision-making quality, protect data and liability, and enhance the effectiveness of the technology investment.
In other words: AI literacy is not a box-ticking exercise. It is the minimum condition for turning a tool into returns.
Three indicators to monitor
If AI is treated as an investment, it must be measured as an investment.
Not the time promised by the vendor, but that observed in the actual process following adoption.
Errors identified, corrections required, consistency of results, reduction in rework.
What people do with the hours freed up: commercial, analytical, decision-making, or relational activities.
The point is not to automate for automation's sake. The point is to increase the organisation's overall return.
The real differentiator will be managerial
Artificial intelligence is a technology, but the competitive advantage it can generate is, above all, managerial.
The businesses that extract value from it will not necessarily be those that adopt the most tools, but those that can do three things better than others: select the right processes, train their people, and measure the return on reallocated time.
For Italian SMEs, this represents a concrete opportunity. The barrier to entry for the technology has fallen. Tools that just a few years ago required substantial investment are now available at far more accessible price points. But for precisely this reason, the advantage will not lie in the tool itself. It will lie in the manner in which it is integrated into the organisation.
Technology can free up time. Only human capital can convert it into value.
This is the true lever of artificial intelligence: not the mechanical reduction of headcount, but the opportunity to increase the productivity of the people already within the organisation.
And, like any lever, it must be applied with method. Because it amplifies results — but also the errors of design.
Sources
- ISTAT, "Enterprises and ICT — Year 2025", press release, December 2025.
- ISTAT, "Enterprises and ICT — Year 2024", press release, 17 January 2025.
- European Union, Regulation (EU) 2024/1689, Article 4 — artificial intelligence literacy.
Disclaimer. This document is for informational purposes only and does not constitute financial advice.
Content produced with the support of artificial intelligence.