Artificial Intelligence in Business: Italy Risks Becoming the Country of Prompts
- Marc Griffith
- May 11
- 5 min read
Summary Artificial intelligence in business: Italian workers are experimenting with AI, but organizations struggle to turn that usage into scalable processes. Key data: 55% are producing work they couldn't handle a year ago; only 10% are Frontier Professionals; organizational leverage outweighs individual initiative. Key takeaways
|
Artificial intelligence in business is now a widespread reality among Italian workers, but the real challenge is not individual adoption: it's transforming personal uses into sustainable organizational processes.
Why the question is changing: from prompts to agents
For months the focus has been on adoption: who uses AI, who doesn't, and for which tasks; today, however, the issue evolves toward designing work around agents, systems capable of performing tasks and coordinating complex workflows.
Companies must rethink processes because it's not enough to leave employees in front of a prompt: we need to transform uses into replicable practices.
Four operating modes with AI
The report identifies four modes of interaction with AI: asking (precise questions), exploration (experimentation), delegation (entrusting structured activities) and collaboration (human–agent co-design).
Understanding when to ask, when to explore, when to delegate and when to collaborate is the practical rule that determines AI's effectiveness at work.
Asking is used for quick and repetitive operations; delegation is the mode that really changes the value of people, shifting it toward direction and control.
What the numbers say: Italy vs. the world
In our country, 55% of AI users report that they are now producing work that a year ago they would not have known how to tackle, and among the most advanced profiles the share rises to 76%.
These figures show that individual experimentation exists, but it is not enough without organizational support.
At the same time, only 10% of Italian users can be classified as Frontier Professionals (versus 16% global average), i.e., those who rethink workflows and build replicable practices.
The gap indicates that often the individual is faster than the company: innovation remains a personal matter rather than a systemic change.
The role of leadership and culture
The Work Trend Index finds that globally only one in four users perceives leadership consistently aligned with AI strategy; in Italy the figure drops to 18%.
If leadership does not align metrics, responsibilities, and incentives, AI remains a sum of micro-efficiencies that are not scalable.
Organizations that create structured conditions for AI use achieve stronger and longer-lasting impacts than those that merely train curious individuals.
What determines real impact
The report distinguishes organizational and individual factors: corporate culture, managerial support, training, and usage rules account for 67% of AI's impact, while individual elements like motivation and familiarity account for 32%.
Investing in processes, rules and managerial training yields more value than relying solely on individual talent.
Emerging skills
Users identify output quality control (50%) and critical thinking (46%) as the most important skills; in Italy values drop to 39% and 36% respectively.
We need to shift emphasis from production to verification: knowing how to assess the quality of an output becomes crucial in the era of agents.
In the era of agents, human value moves toward direction, judgment, and quality control: producing a draft isn't enough; you must be able to evaluate it.
Practical examples of transforming processes
Imagine a team using AI to summarize research: without shared rules, each member will use different tools, with incomparable outcomes and metrics; with shared workflows and templates, the same tool increases quality, replicability, and reduces error risk.
Defining templates, verification criteria, and clear roles is a practical action that enables scaling AI use in the business.
How to measure success
Beyond productivity metrics, companies should introduce indicators of output quality, time spent on checks, and the percentage of processes where AI is an integral part of the workflow.
Measuring the quality of AI-generated output is as important as measuring speed: without proper indicators you won't know what to improve.
The transformation paradox: viewpoints in comparison
On one side are workers who experiment and demand tools: 63% fear falling behind if they don't adapt quickly; on the other, 43% prefer to focus on current goals rather than redesigning processes with AI.
This mismatch shows that individual pressure does not automatically align with an organization's ability to change metrics and responsibilities.
Pro: widespread adoption stimulates rapid innovation and practical skills; access to powerful tools enables results that were previously unimaginable. Con: without governance, operational risks multiply (errors, invented sources, inconsistencies) and opportunities for shared learning are lost.
A balanced approach requires combining free experimentation with operational standards: allowing controlled autonomy and building repeatable practices.
Some managers fear that redesigning work will slow immediate outcomes; others argue that only by changing roles and incentives can AI's full value be unleashed. Implementing pilot programs with clear metrics, dedicated quality-control roles, and incentives for spreading best practices can reconcile the two needs.
Allowing pilot projects with quality KPIs and knowledge-transfer mechanisms is a concrete strategy to overcome the paradox.
Practical recommendations for founders and managers
1) Map critical processes and identify where AI can automate repetitive tasks; 2) Define roles responsible for output quality control; 3) Standardize templates and workflows to make the value produced by individuals replicable; 4) Align KPIs and incentives with the new responsibility for verification.
Applying these four steps reduces the risk that AI remains a sum of micro-efficiencies and enables scaling benefits across the business.
Practical implementation
Launch pilot projects lasting 6–12 weeks, with a measured baseline and targets to improve output quality, to build gradual and measurable roadmaps.
Pilots help validate metrics, identify training gaps, and test control roles before the organizational scale.
Towards new roles and metrics
If an agent can perform an increasing share of activities, human value will shift more toward direction, judgment, quality control, and accountability for the final outcome; it's necessary to rethink job descriptions and evaluation systems.
Reviewing job descriptions and evaluation systems to include verification and judgment skills is an essential operational step.
A pragmatic path to not stay the country of prompts
Italy has ready and curious workers: the challenge is to ensure organizations don't fall behind. Investing in leadership, targeted training, processes, and metrics can transform individual experimentation into sustainable competitive advantage.
Turning individual enthusiasm into business processes is the lever that can transform AI adoption into real and measurable growth.
Final practical notes
Encouraging use alone isn't enough: we need shared criteria, verification roles, templates, and dedicated KPIs. Without these elements AI will continue to be the realm of prompts, not the backbone of organized work.
Building dedicated company criteria and roles is the practical action that allows moving from individual use to systemic adoption.
