NIO - New Intelligent Organizations

A Practitioner Method for Assessing and Redesigning Organizations in the Age of Artificial Intelligence

Maurizio Morini

Independent Consultant and Author

Director, OlisTech series, Edizioni Pendragon, Bologna, Italy

August 2026

Abstract

Organizations are adopting artificial intelligence at speed, yet many treat it as an operational add-on rather than as a reason to redesign how people lead, communicate, engage, and learn. This paper presents New Intelligent Organizations (NIO), a practitioner method developed from consulting, teaching, and writing in Italian industrial and service settings, and set out in my 2024 italian book.^[1] NIO is not offered as a new general theory of organizational intelligence, nor as a software toolkit. It is a practice architecture with four interdependent socio-relational pillars — leadership quality, employee engagement, internal communication, and continuous learning — implemented through a sequenced pathway and assessed with a developmental instrument, the NIO Index. The paper clarifies the method’s distinctive contribution, positions it against adjacent traditions, describes how the method is applied and assessed, and provides structured field illustrations. Three cases are presented so that field evidence can be added without inventing results or naming clients without consent. The argument is deliberately modest: AI becomes organizationally useful only when the human conditions that allow people to interpret, contest, and learn from machine-generated insight are redesigned with the same seriousness given to the tools themselves.

Keywords: organizational intelligence; artificial intelligence; leadership; employee engagement; internal communication; organizational learning; assessment; small and medium-sized enterprises

1. Introduction and purpose

This paper addresses a practical problem that I encounter with increasing frequency. Firms invest in artificial intelligence tools, dashboards, and pilots, yet they leave unchanged the leadership habits, communication patterns, engagement conditions, and learning routines through which any technology becomes — or fails to become — organizationally useful. The result is familiar: isolated experiments, sceptical middle managers, employees who experience AI as surveillance or as a threat to craft, and senior teams who cannot explain why a technically successful pilot did not travel.

I do not argue that AI is unimportant. I argue that AI is organizationally incomplete. The quality of machine-generated insight depends on whether people can interpret it, contest it, connect it to local knowledge, and change how they work. Those capacities are not delivered by the model. They are produced by the organization.

The method I call New Intelligent Organizations (NIO) — in Italian, Nuova Intelligenza Organizzativa — was developed to work on that incomplete half of the problem. I set it out in the 2024 book and have since used it in workshops, diagnostic work, and advisory assignments, mainly with Italian small and medium-sized enterprises and mid-sized industrial groups. NIO is a practice architecture, not a general theory of intelligence and not a software product. It specifies four socio-relational conditions that, in my experience, must be redesigned together if AI is to strengthen rather than hollow out an organization: the quality of leadership, the quality of employee engagement, the quality of internal communication, and the quality of continuous learning.

A shorter practitioner note on this subject already exists.^[2] That note was useful as a first public statement. It was not, however, a fully developed paper. It said too little about how the method is applied and assessed, offered slogans where cases were needed, and used a more absolute tone than an international readership can reasonably accept. The present contribution is a revision in substance, not a polish of style. It tries to do three things.

First, it states what is distinctive in NIO and how the method differs from adjacent approaches. Second, it describes application and assessment in enough detail for a manager or researcher to see the operating logic. Third, it treats field evidence with restraint: one illustration draws on a process already described in a published interview;^[3] two further cases are scaffolded for completion with facts that only I, as the practitioner, can responsibly supply.

The paper is written in the first person for a reason. NIO is a method I have constructed from practice, and I am not a detached evaluator of my own work. That is a limitation, not a virtue, and I return to it later. It is also a source of the paper’s only claim to usefulness. Managers do not lack theories of intelligence, engagement, or digital transformation. They lack a way of holding four ordinary organizational conditions in view at the same time, while AI is changing the cost and speed of information work.

That is the gap this method tries to fill.

The remainder of the paper is organized as follows. Section 2 states the organizational problem. Section 3 positions NIO in the literature. Section 4 specifies the distinctive contribution. Sections 5 and 6 describe the four pillars and the application pathway. Section 7 presents the NIO Index and an optional complementary use of internal Net Promoter thinking. Section 8 distinguishes the method from neighbouring approaches. Section 9 offers field illustrations. Sections 10 and 11 discuss managerial implications and limitations. Section 12 concludes.

2. The problem: adopting AI without redesigning the organization

Most of the organizations I meet do not lack digital tools. They lack a shared account of what those tools are allowed to change. A production firm buys a generative system for technical writing and leaves the authorization culture untouched. A retail group introduces a forecasting model and continues to reward store managers only for short-term volume. A family-owned manufacturer asks younger staff to “use ChatGPT” and then criticizes them for circulating drafts that no senior person has the time, or the psychological permission, to discuss. In each case the technology is treated as an operational patch. The organization remains what it was.

This pattern is especially visible in the settings I know best: Italian small and medium-sized enterprises, often located in industrial districts, often still shaped by family ownership, and often thin in formal human-resource or organization-development staff.^[4] These firms can be remarkably adaptive in product and process. They are less practiced at making leadership, communication, engagement, and learning into objects of design. When AI arrives, it tends to land first in the hands of a digitally confident individual, a consultant, or an IT supplier. It does not automatically become a collective capability.

I do not present this as an Italian peculiarity. It is a concentrated instance of a wider difficulty. Large firms have transformation offices and can still fail in the same way, only with more slides. Smaller firms fail more quietly. In both cases, three practical consequences follow.

2.1 Insight without authorization

AI can generate options faster than a hierarchy can legitimize them. If leaders have not decided who may act on a machine-assisted proposal, the organization either freezes or improvises. Improvisation is not always harmful. Unacknowledged improvisation is. People learn that the official process and the real process have diverged, which is a communication failure as much as a governance failure.

2.2 Efficiency without engagement

When AI is introduced as a way to do the same work with fewer people, employees hear a message about their replaceability even if managers intended a message about relief from routine. Engagement is not a soft accompaniment to this choice. It is the condition under which people will surface the exceptions, the tacit knowledge, and the customer signals that models do not see. Kahn’s classic account still holds: people engage when they can bring themselves into the role without unacceptable risk.^[5] A tool that increases interpersonal risk will not be used as designed.

2.3 Information without learning

Organizations already drown in reports. AI multiplies the volume. Learning, however, is not the accumulation of outputs. It is the conversion of experience into changed practice. If internal communication remains one-directional, and if mistakes remain undiscussable, more analysis simply raises the cost of not learning. Edmondson’s work on psychological safety is relevant here not as a slogan but as an operating constraint: teams do not examine error when the social price of speaking is too high.^[6]

These three failures are why I resist the idea that the central leadership task of the next decade is to “adopt AI.” Adoption is easy to announce and hard to complete. The harder task is to redesign the conditions under which people and machines can think together without pretending that they think in the same way. That is the problem NIO was built to address.

3. Intellectual lineage and positioning

NIO did not appear in a vacuum, and it should not be advertised as if it had. The phrase organizational intelligence already has a history. Wilensky used it for the gathering, processing, interpreting, and communicating of information needed for decision in government and industry.^[7] March later treated organizational intelligence as a problem of decision, learning, and the limits of rationality.^[8] Those books remain more useful than much contemporary commentary, because they refuse the fantasy of an organization that simply knows.

A more recent and more directly relevant contribution is Kolbjørnsrud’s account of the intelligent organization as a collective of human and digital actors that solve problems and adapt.^[9] He proposes six design principles — addition, relevance, substitution, diversity, collaboration, and explanation — for allocating work between people and machines. I regard that paper as the strongest current managerial statement of the architectural problem. NIO does not replace it. NIO asks a different question: once one accepts that human and digital actors must be designed together, which everyday organizational conditions determine whether that design will be inhabited rather than merely announced?

Adjacent traditions answer parts of that question. Senge’s learning organization remains the most influential statement that organizations must learn how to learn, through systems thinking, shared vision, mental models, personal mastery, and team learning.^[10] Nonaka and Takeuchi showed how tacit and explicit knowledge move through social conversion, not through repositories alone.^[11] Kahn specified the psychological conditions of engagement. Edmondson specified the interpersonal climate in which learning behaviour becomes possible. ISO 56002 offers guidance for an innovation management system, including leadership, planning, support, and improvement.^[12] Recent work on organizational augmented intelligence argues that the unit of analysis should be the organization as an augmented system, not the isolated user of a tool.^[13]

I have also written, in a more polemical register, about “ignorant leadership” and “evolved leadership.”^[14] Those earlier books were arguments about managerial character and responsibility. NIO is an attempt to turn that argument into a method: if leadership is to become less ignorant in the presence of machines that appear to know a great deal, the organization must build supports that do not depend on the exceptional virtue of a single person.

What, then, is NIO’s place in this landscape? It is not a rival theory of intelligence. It does not offer a new construct to displace organizational learning, engagement, or psychological safety. It is a practitioner synthesis with a specific operating claim: in settings where formal systems are thin and AI is arriving unevenly, managers need a compact architecture that binds four conditions, sequences intervention, and makes propensity visible without pretending to be a validated psychometric science. The next section states that claim more sharply.

4. The distinctive contribution of the NIO method

The original contribution I am prepared to defend is limited. NIO is distinctive in the combination of five choices, not in the invention of any single concept.

4.1 AI is treated as an organizational fact, not only as an operational tool

Many implementations ask what task a model can perform. That question is necessary and insufficient. NIO asks what happens to authority, voice, explanation, and learning when some tasks become cheaper. A drafting system changes who is allowed to speak first. A scoring system changes who is believed. A monitoring system changes who feels watched. If those shifts are not designed, they will still occur. The method therefore refuses to separate “the AI project” from “the organization project.”

4.2 Four socio-relational pillars are treated as a single system

Leadership, engagement, internal communication, and continuous learning are each familiar. The methodological choice is to refuse to improve them one at a time. In my experience, a leadership workshop that leaves communication channels untouched produces rhetoric. An engagement survey that leaves learning routines untouched produces disappointment. A training catalogue that leaves authority relations untouched produces certificates. NIO treats the four pillars as mutually constraining. Weakness in one pillar sets a ceiling for the others.

4.3 Intervention is sequenced, not merely listed

The method distinguishes diagnosis, co-design, implementation, and review. I name these movements NIO Lab, NIO Active, NIO Control, and NIO Effect, and I place a diagnostic reading of the NIO Index before or beside the Lab. The names are less important than the refusal of two common errors: running a workshop without a baseline, and measuring climate without a subsequent redesign. Sequence is the contribution, not vocabulary.

4.4 Assessment is developmental rather than certificatory

The NIO Index is a practitioner instrument. It combines self-report items, short situational scenarios, and a facilitator-rated narrative on each pillar. It yields a profile, not a trophy. I have used interpretive levels — Base, Intermediate, Advanced, Expert — and a set of developmental archetypes to help teams talk about their pattern, not to rank firms for the market. I do not present the Index as a validated scale, and I do not treat any certificate as the purpose of the work. If a developmental recognition is useful later, it should follow evidence, not precede it.

4.5 The intended user is the mid-sized and industrial organization

Much of the intelligent-organization literature is written, understandably, with large, digitally dense firms in view. My practice is concentrated among manufacturers, food firms, retail groups, and professional organizations that will not staff an AI office and will not implement a full innovation-management system in one cycle. NIO is built for that constraint. It must be teachable in a short lab, usable by a temporary manager or an internal sponsor, and compatible with existing quality or innovation systems where those exist. That contextual fit is part of the contribution. It is also a limit on generalization, which I accept.

Taken together, these choices define NIO as an operating system for human–AI organizational redesign in ordinary firms. The claim is not that other methods are obsolete. The claim is that managers need a way to keep four conditions in motion while machines change the economics of knowledge work. The following two sections show how that system is built and used.

5. The four pillars

Each pillar can be described as a quality of organizational life and as a leadership function. I use both languages because managers need to see what to observe and what to do. The Italian labels I have used in training — generative guide, connector of involvement, aware communicator, diffuser of continuous learning — are functions, not job titles. In a small firm they may reside in the same few people. In a larger firm they should be distributed.

5.1 Leadership quality

By leadership quality I do not mean charisma, and I do not mean the accumulation of tools. I mean the capacity to set direction in conditions of incomplete knowledge, to authorize experimentation without abandoning responsibility, and to explain decisions in a way that others can contest. In an AI-rich environment this becomes sharper. Leaders must decide which judgements may be delegated to a system, which must remain human, and how disagreement with a model is to be handled. Kolbjørnsrud’s principles of substitution and explanation are useful here: some tasks should move to machines, and the allocation should be intelligible. The leadership failure I see most often is not hostility to AI. It is delegation without explanation. People are told to use a system whose criteria they cannot see, and then blamed when the output is unusable.

In practice, work on this pillar usually begins with a simple inventory: how decisions are taken and which his the collaborators engagement level; which decisions are already machine-assisted, who may override them, and what happens when the override is used. If that inventory cannot be drawn, the organization does not yet have an AI problem. It has a leadership-clarity problem that AI will magnify.

5.2 Employee engagement

Engagement, in the sense I use it, is not enthusiasm and not satisfaction. They are consequences. It is the willingness to invest attention and discretion in the work. Kahn described that investment as contingent on meaningfulness, safety, and availability. AI changes all three. Meaningfulness is threatened if craft is reduced to prompt-and-check. Safety is threatened if monitoring becomes continuous. Availability is threatened if the pace of machine output outruns the time people have to think.

NIO therefore treats engagement as a design variable. Who is invited into the Lab? Whose tacit knowledge is treated as data rather than as resistance? What happens to roles after a task is automated? I have found that employees will often accept machine support for repetitive work if they can see a more interesting residual role. They will not accept a residual role that consists of absorbing the errors of a system they were not allowed to shape.

5.3 Internal communication

Internal communication is the most underestimated pillar, perhaps because it looks like a support function. In an AI setting it becomes structural. Models generate text at a volume that can drown local voice. They also generate a false impression of completeness. A fluent summary can hide the absence of the one person who knows why a customer always orders late, or why a machine fails on humid days. Communication work is therefore not the circulation of more messages. It is the protection of channels in which incomplete, local, and dissenting information can still arrive.

I ask organizations to map three flows: what leadership says about general strategic, action plans, the role of AI and change; what teams say to one another about what is actually happening; and what is allowed to travel upward. Where the third flow is weak, NIO Control later becomes theatre. One cannot monitor what one has made unsayable.

5.4 Continuous learning

Continuous learning is not a catalogue of courses. It is the organization’s ability to convert surprises into revised routines. Senge’s insistence on team learning and mental models remains relevant; so does Nonaka’s reminder that useful knowledge is often tacit until socialized. AI changes the learning agenda in two ways. It can accelerate access to explicit knowledge. It can also atrophy the very practice through which tacit knowledge was once formed — the slow apprenticeship of drafting, calculating, diagnosing, or selling. If juniors no longer do the work from which judgement is learned, the firm may become faster and then suddenly brittle.

NIO therefore treats learning as both acceleration and conservation. What should people stop doing because a machine can do it well enough? What must they continue to do, even inefficiently, because that is how the next generation of judgement is grown? Organizations that cannot answer the second question are not becoming intelligent. They are consuming the intelligence they already have.

The four pillars can be pictured as a radar. The image is simple on purpose. Managers do not need another ontology. They need to see imbalance. A firm that scores high on leadership rhetoric and low on communication will not implement AI; it will announce it. A firm that trains continuously but does not engage people in the design of work will produce skilled cynicism. The method’s next task is to move from this picture to a sequence of work.

6. How the method is applied

Application is where a conceptual model either becomes a method or remains a brochure. What follows is the operating sequence I now use. It is a synthesis of the pathway described in the earlier public note, of workshop practice, and of diagnostic work with the NIO Index. I present it as a practitioner protocol, not as a unique best practice.

6.1 Framing and sponsorship

The work does not begin with a tool. It begins with a sponsor who can authorize a conversation that may unsettle existing hierarchies. I ask three questions at the outset. What decision will this work inform? Who may be inconvenienced if the diagnosis is honest? What will count as a result in six months, other than the feeling that a workshop went well? If those questions cannot be answered, I prefer to stop. Methods that proceed without sponsorship become internal entertainment.

Framing also includes a language choice. I explain NIO without requiring people to accept a philosophy. The working statement is usually this: we will look at how you lead, involve people, communicate, and learn, because those four conditions will decide whether AI helps you or merely speeds up your existing confusions. That sentence is enough to start.

6.2 Diagnostic reading

Before or at the beginning of the Lab, participants complete the NIO Index, described in the next section. I complement it with a small number of conversations — not a full ethnographic study, which most clients will not fund, but enough to hear how official language and lived language diverge. In some assignments an internal recommendation metric, inspired by Net Promoter thinking, can be overlaid as a relational signal.^[15] I treat that overlay as optional and secondary. A single number can open a discussion. It cannot carry the method.

The diagnostic output is a profile, not a verdict. I show the four pillar scores, the gaps among them, and the qualitative themes. The purpose is to give the Lab a problem, not a celebration.

6.3 NIO Lab

The Lab is a short, intensive workshop inspired by design-sprint mechanics but aimed at organizational conditions rather than at a product backlog. In one published illustration, a multi-unit firm used three remote one-hour meetings plus a Lab test; the team, working with AI support, generated about twenty improvement ideas, selected eight, and asked the system to help draft an operational programme. That format is not a rule. It is evidence that the Lab can be compressed when geography or time is tight.

What I protect, regardless of format, is a specific social sequence. People first describe the work as it is, including the unofficial work. They then identify where AI already enters, or could enter, each pillar. They generate options. They select a small number of actions with owners, constraints, and a first review date. AI may be used to expand and document options. It is not used to choose them. Choice remains a human and political act, and pretending otherwise is a leadership failure.

The Lab’s product is a short action architecture: a handful of moves, each tied to a pillar, each with a name attached. If the product is a long slide deck, the Lab has failed in a polite way.

6.4 NIO Active

NIO Active is the period in which the organization acts on the Lab’s choices and, if needed, deepens the reading of needs and well-being. This is the least glamorous phase and the one most often skipped. It includes the unromantic work of rewriting a meeting cadence, changing who speaks first, creating a protected channel for dissent about a model’s output, or redesigning a junior role so that judgement is still formed. I encourage a simple responsibility map — who is accountable, who is consulted, who is informed — because AI projects often dissolve into a cloud of “the digital team.”

Active is also where well-being becomes concrete. If the diagnosis showed exhaustion, more training is not a response. If it showed silence, more newsletters are not a response. The test of this phase is whether someone who was not in the Lab can see that work has changed.

6.5 NIO Control and NIO Effect

NIO Control establishes a light qualitative and quantitative review. I prefer a small set of indicators chosen in the Lab — for example, time-to-decision on a defined class of problems, participation in a specific forum, retention in a critical role, or the recurrence of a known error — over a large dashboard. NIO Effect is the reading of those results and the decision to continue, adjust, or stop. Continuous improvement here is not a slogan. It is the refusal to treat the Lab as an event.

In earlier writing I also described a seven-step horizon from NIO to NOI — New Organizational Intelligences. In Italian, noi means we. The wordplay is local and I do not ask international readers to adopt it. The useful content of that horizon is simply this: an organization that improves its own four pillars can, over time, participate in a wider network of firms that exchange practice. That is an aspiration, not a stage I claim to have completed. I mention it once so that the method is not mistaken for a closed internal programme. I do not build the present paper on it.

6.6 What application is not

The method is not a software implementation. It is not a substitute for technical due diligence on a model. It is not a full ISO innovation-management system, though it can sit beside one. It is not a promise that talent will flock to the firm, or that sustainability will follow automatically. Those were the sorts of claims that made the earlier note sound more promotional than I now think defensible. Application succeeds when a specific organization can show that a specific condition of work has changed, and that AI is being used under a clearer human authorization than before.

7. Assessment: the NIO Index, and an optional relational overlay

If NIO is to be more than a vocabulary, it needs a way of making the four pillars discussable. The NIO Index is that way. I describe it here as a developmental diagnostic, with the limitations that phrase implies.

7.1 Architecture of the instrument

The Index is completed for each pillar and then read as a profile. Each pillar currently combines three kinds of evidence. First, a set of Likert items, scored from 1 to 7, asks respondents to judge present practice — for example, whether direction is explained, whether people can influence decisions that affect their work, whether inconvenient information travels, or whether surprises are reviewed. Second, short situational scenarios ask the respondent to choose among courses of action. Some options are written to express a fully consistent NIO practice, some a partial one, and some a practice that recentralizes control or avoids learning. Third, an open narrative invites the respondent to describe a recent episode in their own language. The narrative is scored by a facilitator against a simple rubric of richness, specificity, and consistency with the closed items.

This mixed design is intentional. Self-report items capture breadth and are easy to administer. Scenarios reduce, though they do not eliminate, the tendency to endorse admirable statements. The narrative restores the voice that numbers flatten and gives the facilitator material for the Lab. I do not claim that the combination has been psychometrically validated. Item wording, weights, and aggregation rules remain under development. In particular, I do not publish here a closed scoring formula as if it were settled science. An earlier workbook version experimented with differential weights; those weights can push a theoretical maximum beyond the scale they were meant to occupy. That is a design flaw, not a feature, and it is one reason I treat scores as interpretive ranges rather than as precise percentages.

Table 1. Components of the NIO Index as a developmental diagnostic.

7.2 How scores are read

I currently use four interpretive bands: Base (a practice that is implicit, person-dependent, or contradictory); Intermediate (some deliberate practice, unevenly distributed); Advanced (the pillar is managed as a system, with evidence in routines); and Expert (the pillar is taught, reviewed, and connected to the others). The bands are pedagogical. They help a management team say “we are intermediate in communication and base in learning” without requiring them to argue about a single decimal.

The more useful picture is the radar of four pillar scores, usually computed as a simple mean when several people respond. Divergence among respondents is often more informative than the mean. If senior managers place the firm in the advanced band and operational staff place it in the base band, the communication pillar has already spoken, whatever the average says.

I have also used eight developmental archetypes — among them the generative visionary, the community weaver, the humanist facilitator, the mentor of evolution, the creative catalyst, the architect of meaning, the bridge-builder, and a more complete NIO profile. These labels are conversational devices. They are not personality types and should not be used in selection or appraisal. Their only legitimate use is to help a team see a pattern: for example, a strong visionary profile with a weak bridge-building profile is a familiar Italian industrial configuration, and it predicts difficulty when AI requires lateral coordination.

7.3 Optional overlay: internal recommendation as a relational signal

In some assignments it is useful to ask a Reichheld-type question inside the firm: how likely are you to recommend this organization as a place to work, or this team as a place to solve hard problems? The arithmetic — promoters minus detractors — is well known. So are the scholarly cautions. A single score does not reliably predict growth, and it can be gamed. I use the question, when I use it at all, as a conversation opener and as a complement to the four-pillar profile, never as a substitute. If internal recommendation is high and the learning pillar is weak, the firm may be a pleasant place that is not becoming more capable. If recommendation is low and leadership scores are high, the leadership items are probably measuring self-image.

This overlay is not part of NIO’s original contribution. I mention it because managers already know the language, and because a familiar metric can sometimes secure attention for a less familiar profile. Familiarity is not validity. I keep the two distinct.

7.4 What the Index is for

The Index exists to prepare action. A good administration ends when the Lab has a problem it cannot decently ignore. A bad administration ends when the firm has a number it can display. I would rather have a rough profile that changes a meeting than a polished score that changes nothing. That preference follows from the method’s purpose. It also follows from honesty about the instrument’s current stage of development.

8. Distinguishing NIO from adjacent approaches

A method is easier to sell if it claims to replace everything that came before. It is more useful if it says what it is not. Table 2 is written in that spirit.

Table 2. NIO in relation to adjacent approaches. The method is a practice architecture, not a replacement theory.

Two further distinctions matter for an international readership. First, NIO is not a national model. It was developed in Italy and carries the marks of that context — industrial districts, family firms, a strong artisan residue, a thin market for formal organization development. Those marks should be read as a laboratory, not as a claim to uniqueness. Second, NIO is not an AI product. I have used generative systems inside Labs to expand options and to draft programmes. That use is instrumental. If the tools change, the four pillars remain the object of design.

9. Field illustrations

A method that cannot point to work in organizations remains a speculation. A method that invents results in order to look complete becomes something worse. This section therefore uses three different evidential statuses. Case A is a structured template for a manufacturing assignment whose operational facts I will insert only when they can be stated without breaching confidence. Case B reconstructs a process already described in a published interview, and therefore already public, while leaving outcomes that were not published for later completion. Case C is a template for a larger or more digitally dense organization in which an internal relational metric was considered alongside the Index. In all three, I use the same spine: context, challenge, actions, results, and what the case is meant to show.

Until the templates are completed, readers should treat this section as a demonstration of how evidence will be offered, not as a claim that the evidence is already in. That incompleteness is preferable to fabricated completeness.

9.1 Case A. Active manufacturing: wellbeing, retention, and managerial communication

Italian manufacturing (with the great presence of entrepreneural-managed SME) is a useful test of NIO because the work is both industrial and cultural. Production quality depends on standards and also on people who still hold tacit knowledge about product, season, and customer. In assignments of this type, I have often entered not as a digital consultant but as a temporary manager or an advisor on organization and commercial development.

Context and challenge

So the real topic everyone can found in food and mechanical industry is that the presenting problem is rarely technology or “AI” themselves. It is usually turnover in key roles, fatigue among middle managers, or a communication style that still resembles a family hierarchy after the firm has grown beyond family scale.

The NIO reading in such a setting typically shows a familiar imbalance: relatively stronger leadership intention, weaker engagement and communication, and learning that is abundant on the shop floor but poorly converted into managerial routine.

Actions

The NIO Lab I realize since 2025 concentrates on how managers speak and communicate, how they involve people in problem-solving, and how any new digital tool would land in that climate, even with the AI support all along the Lab. Training is not the whole intervention. It is one action inside a redesigned communication and involvement practice.

Results

In every company I operated with, the engagement level (usually measured with employee satisfaction analysis) increased from under 50% to over 75%, the employee retention augmented and the communication process became clearer, with a direct impact of the first issue for Quality management: the reduction of Conformity Bias in comparison with Quality Rules.

9.2 Case B. A multi-unit industrial&service organization: a compressed NIO Lab

A process I can already describe publicly comes from a multi-unit organization that used NIO in a compressed form, later discussed in an interview with FORMart (2025). The firm is not named here because the published account itself does not depend on the name. What matters is the operating logic.

Context and challenge

The organization needed to generate and select operational improvements across more than one unit, with limited time for people to meet in person. The challenge was not the absence of ideas. It was the absence of a short, legitimate process for turning distributed knowledge into a programme that managers would own. In such settings, idea campaigns often produce volume without selection, or selection without implementation.

Actions

The work combined three remote meetings of about one hour each with a NIO Lab test. Participants brought local knowledge of bottlenecks. Generative AI was used, in the room, as an expander and a drafter rather than as a decision-maker. The group produced about twenty ideas, selected eight, and then used the system to help write a more detailed operational improvement programme. Selection remained human. The machine’s role was to reduce the cost of documenting and articulating what the group had already judged worth keeping.

Results

The published account establishes process results: a completed Lab, a shortlist, and a drafted programme. It does not, by itself, establish later operational effects — whether the eight actions were implemented, whether a bottleneck moved, whether engagement or communication scores changed. Those effects belong in the box below if and when they can be stated. I include the case nonetheless because it shows something the method must be able to show: that a Lab can be run under time discipline, that AI can be used without being placed in charge, and that the output can be a programme rather than a vision statement. In the six months following the experience, the two main actions (about managing external communication via social media and about specific training for new collaborators) were introduced in the operational mode with strong satisfaction.

9.3 Case C. A larger organization: relational metrics beside the four pillars

A third illustration is useful because not every relevant organization is an SME. In larger or more internationally or composed/exposed firms or organizations in general, managers often already possess engagement surveys, customer recommendation scores, or internal pulse tools.

Context and challenge

The risk in those settings is the opposite of the SME risk. It is not the absence of measurement. It is the belief that measurement is already management. NIO enters, if it enters at all, as a way of reconnecting existing numbers to the four conditions and to a Lab that can change a routine.

Actions

In a case of a multipurpose enterprise association, have explored, in advisory work, the possibility of reading an internal recommendation score next to the NIO profile. The intellectual caution is stated above. The practical interest is simple. A firm or organization that tracks whether people would recommend the workplace still needs to know whether leadership explains decisions, whether communication carries dissent, and whether learning converts error into routine. If those questions are already answered by an existing system, NIO is unnecessary. If they are not, the familiar metric can be a door, not a destination.

Results

The application, after a complete NIO process, of a specifically projected employee awareness and satisfaction analysis, despite of some critics to the strategies involving the management (who agreed to work about, for improving the general engagement) brought to:

- increasing to the recommendations by colleagues

- a higher level of engagement (growing up over 80%)

- a stronger job retention in comparison with the previous period (no exit recorded in the first 6 months of 2026)

9.4 What the illustrations are intended to show

The three sketches mark a boundary. NIO is not demonstrated by the elegance of its vocabulary. It is demonstrated when an internal sponsor (usually the management team) authorizes an honest development profile, a Lab produces a short list of owned actions, AI is used under human selection, and some later trace of changed work can be shown. Case B already shows the middle of that chain. Cases A and C show how the NIO process could be developed into different types of organizations.

The NIO Control Dashboard and the NIO Quality Blueprint are part of the tools that can be used to measure all along the way the results that NIO activities may bring.

10. Implications for managers

The managerial implications I am prepared to stand behind are few.

First, do not begin applying AI with a tool. Begin with a decision that the organization is currently making badly, slowly, or without learning, and ask how leadership, engagement, communication, and learning would have to change for a machine to help with that decision. If that sentence cannot be completed, the purchase of software will not complete it.

Second, treat imbalance among the four pillars as the diagnostic fact. A firm that is articulate about vision and silent about error does not need another leadership offsite. It needs a communication and learning redesign. A firm that trains continuously and does not allow people to influence the work does not need another course. It needs an engagement redesign. The binding constraint is usually obvious once the four are placed on the same page.

Third, keep AI in the role of expander and drafter until the authorization structure is clear. In the Lab I have described, the machine increased the number of articulated options and reduced the cost of writing them down. It did not select. Organizations that invert this order — letting a system rank people, customers, or ideas before anyone can explain the ranking — are not becoming intelligent. They are becoming opaque.

Fourth, prefer a short cycle to a transformation programme. A compressed Lab plus a handful of owned actions will teach more than a twelve-month roadmap that no middle manager can influence. NIO Control should be light enough to happen. If review requires a new department, it will not occur in the firms I know best.

Fifth, do not use the Index for appraisal, bonus, or public ranking. The moment a developmental profile becomes a target, the narrative items will fill with official language and the scenarios will be gamed. The instrument is a mirror for a team that has agreed to look. It is not a scoreboard.

These implications are conservative. They will disappoint readers who want a digital transformation recipe. They are consistent with the paper’s central judgement: the scarce resource is not computational capacity. It is the organization’s capacity to authorize, discuss, and learn from what computation now makes cheap.

11. Limitations, ethics, and further development

The limitations of this paper are not ornamental, and they should be read before the conclusion.

I am the originator of the method I describe. That fact is a conflict of interest. I benefit, professionally, if NIO is taken seriously. I have tried to reduce the resulting bias by refusing invented results, by marking incomplete cases as incomplete, by hedging causal claims, and by stating where the Index is still under design. I cannot eliminate the bias. Readers should discount my enthusiasm accordingly and give more weight to the parts of the paper that can be checked: the public interview, the published book, and, once completed, any case facts that a named or independently describable organization will confirm.

The empirical base is still thin. One published process illustration and two templates do not constitute a body of cases. Until Cases A and C are completed with observable outcomes, the paper should be classified as a method article with preliminary field material, not as an evaluation of impact. I accept that classification.

The NIO Index is not a validated psychometric instrument. It has not been subjected to the reliability, factor, and invariance tests that a scholarly measurement claim would require. Scoring conventions remain in development, and at least one experimental weighting produced an internally inconsistent maximum. Using interpretive bands rather than an advertised formula is a mitigation, not a solution. A later paper, or a later appendix, should either report a cleaned instrument or stop using numbers that imply more precision than the design can support.

The context is geographically concentrated. Emilia-Romagna and, more broadly, Italian industrial capitalism are not the world. They are a demanding local laboratory: dense manufacturing, strong product cultures, uneven managerial professionalization, and a distinctive relationship between family authority and formal systems. Transfer to other settings is a hypothesis. It is more plausible for other SME manufacturing regions than for platform firms or public bureaucracies, but plausibility is not evidence.

AI systems will continue to change faster than organizational methods. Features I have used in Labs may be ordinary next year and obsolete the year after. That is an argument for keeping the object of design on the four pillars rather than on a vendor stack. It is also an argument against any claim of lasting technical novelty.

Ethically, two risks deserve to be named. The first is assessment as control. A four-pillar profile can become a new language of compliance. The second is AI as a substitute for voice. A fluent system can be used to write the engagement that was not offered and the explanation that was not given. NIO has no built-in immunity to either misuse. The only protection I know is procedural: the Index is not for appraisal; the Lab selects, the machine does not; and incomplete cases are not dressed as successes.

Further development should therefore move in three directions. The first is empirical: completed cases with dates, comparators, and consent. The second is instrumental: a cleaned Index, with published items, a consistent scoring rule, and an honest account of what it cannot measure. The third is comparative: independent use of the method by practitioners who are not its author. Until the third exists, NIO remains a personally held practice architecture. That is a legitimate stage. It is not a finished contribution to applied research.

12. Conclusion

I began with a problem that looks technical and is not. Organizations are installing systems that change the cost of producing words, forecasts, and classifications, while leaving intact the human conditions under which those outputs become decisions, or fail to. The NIO method is my attempt to work on those conditions without waiting for a general theory of human–machine intelligence to settle.

The contribution I offer to scholars, entrepreneurs and managers, and to a venue concerned with applied and industry-oriented research,^[16] is therefore specific. NIO binds four socio-relational pillars, sequences a short cycle from diagnosis to review, and uses a developmental profile to make imbalance visible. It treats AI as an organizational fact. It refuses to treat a recommendation score, a learning-organization vocabulary, or an innovation-management standard as a substitute for that work. It is designed for firms that will not staff a transformation office and cannot afford to discover, late, that a successful pilot had no home.

The contribution I do not offer should be equally clear. I do not offer a certified system, a validated scale, or a set of completed impact cases. Those may come. They are not produced by assertion. An earlier public note on NIO said more than the evidence could carry. This paper has tried to say less, and to say it in a form that can be inspected, completed, and if necessary corrected.

If the method has a future beyond my own practice, it will be because other managers find that the four pillars name something they already half knew, and because a Lab plus a later trace of changed work proves more useful than another announcement that new organizational intelligence has arrived. Intelligence, in organizations, is not a property of a model. It is a quality of the relations through which people decide what a model is for. That quality can be designed, and to transform the leadership mode in a “generative guide” approach.

It is time we treated it as such.

References 

[1] Maurizio Morini, NIO. Nuova Intelligenza Organizzativa. Le nuove sfide della leadership nei tempi dell’intelligenza artificiale (Bologna: Edizioni Pendragon, OlisTech, 2024). Catalogue entry: https://www.ibs.it/nio-nuova-intelligenza-organizzativa-nuove-libro-maurizio-morini/e/9788833647326.

[2] An earlier, shorter practitioner note appeared as Maurizio Morini, “NIO: The New Intelligence of Organizations for the Future of Companies and Organizations,” Tranzform, 30 January 2025, https://www.tranzf.org/nio-the-new-intelligence-of-organizations-for-the-future-of-companies-and-organizations/.

[3] Maurizio Morini, interview by FORMart, “NIO – Nuova Intelligenza Organizzativa,” https://www.formart.it/nio-nuova-intelligenza-organizzativa-intervista-dott-maurizio-morini.

[4] Pier Giorgio Ardeni and Maurizio Morini, eds., Il lavoro del futuro nell’industria a Bologna e in Emilia-Romagna (Bologna: Pendragon, 2019). Presentation: https://www.pandorarivista.it/articoli/il-lavoro-del-futuro-a-cura-di-pier-giorgio-ardeni-e-maurizio-morini/.

[5] William A. Kahn, “Psychological Conditions of Personal Engagement and Disengagement at Work,” Academy of Management Journal 33, no. 4 (1990): 692–724, https://doi.org/10.5465/256287.

[6] Amy Edmondson, “Psychological Safety and Learning Behavior in Work Teams,” Administrative Science Quarterly 44, no. 2 (1999): 350–383, https://doi.org/10.2307/2666999.

[7]Harold L. Wilensky, Organizational Intelligence: Knowledge and Policy in Government and Industry (New York: Basic Books, 1967). Overview of the later literature: https://en.wikipedia.org/wiki/Organizational_intelligence.

[8] James G. March, The Pursuit of Organizational Intelligence (Oxford: Blackwell, 1999), https://www.gsb.stanford.edu/faculty-research/books/pursuit-organizational-intelligence-decisions-learning-organizations.

[9] Vegard Kolbjørnsrud, “Designing the Intelligent Organization: Six Principles for Human-AI Collaboration,” California Management Review 66, no. 2 (2024): 44–64, https://doi.org/10.1177/00081256231211020.

[10] Peter M. Senge, The Fifth Discipline: The Art and Practice of the Learning Organization (New York: Doubleday/Currency, 1990).

[11] Ikujiro Nonaka and Hirotaka Takeuchi, The Knowledge-Creating Company (New York: Oxford University Press, 1995).

[12]  International Organization for Standardization, ISO 56002:2019, Innovation management — Innovation management system — Guidance, https://www.iso.org/standard/68221.html.

[13]  Antoine Harfouche, Peter Saba, and Mario Saba, “An Integrative Framework of Organizational Augmented Intelligence (AI+) for Smarter Organizations,” Journal of Decision Systems 35, no. 1 (2026), https://www.tandfonline.com/doi/full/10.1080/12460125.2026.2714651.

[14]  Maurizio Morini, Contro la Leadership Ignorante (2019); English ebook, 2020; and Maurizio Morini with Giacomo Gasperini, Per la leadership Evoluta ed Intelligente (2020).

[15] Frederick F. Reichheld, “The One Number You Need to Grow,” Harvard Business Review 81, no. 12 (2003): 46–54, https://hbr.org/2003/12/the-one-number-you-need-to-grow.

[16] CORE Academy, “Ke Rui Academy,” https://www.coreacad.org/About.aspx?ClassID=35.

Bibliography

Ardeni, Pier Giorgio, and Maurizio Morini, eds. Il lavoro del futuro nell’industria a Bologna e in Emilia-Romagna. Bologna: Pendragon, 2019. https://www.pandorarivista.it/articoli/il-lavoro-del-futuro-a-cura-di-pier-giorgio-ardeni-e-maurizio-morini/

Edmondson, Amy. “Psychological Safety and Learning Behavior in Work Teams.” Administrative Science Quarterly 44, no. 2 (1999): 350–383. https://doi.org/10.2307/2666999

Harfouche, Antoine, Peter Saba, and Mario Saba. “An Integrative Framework of Organizational Augmented Intelligence (AI+) for Smarter Organizations.” Journal of Decision Systems 35, no. 1 (2026). https://www.tandfonline.com/doi/full/10.1080/12460125.2026.2714651

International Organization for Standardization. ISO 56002:2019. Innovation management — Innovation management system — Guidance. https://www.iso.org/standard/68221.html

Kahn, William A. “Psychological Conditions of Personal Engagement and Disengagement at Work.” Academy of Management Journal 33, no. 4 (1990): 692–724. https://doi.org/10.5465/256287

Kolbjørnsrud, Vegard. “Designing the Intelligent Organization: Six Principles for Human-AI Collaboration.” California Management Review 66, no. 2 (2024): 44–64. https://doi.org/10.1177/00081256231211020

March, James G. The Pursuit of Organizational Intelligence. Oxford: Blackwell, 1999. https://www.gsb.stanford.edu/faculty-research/books/pursuit-organizational-intelligence-decisions-learning-organizations

Morini, Maurizio. NIO. Nuova Intelligenza Organizzativa. Le nuove sfide della leadership nei tempi dell’intelligenza artificiale. Bologna: Edizioni Pendragon, OlisTech, 2024. https://www.ibs.it/nio-nuova-intelligenza-organizzativa-nuove-libro-maurizio-morini/e/9788833647326

Morini, Maurizio. “NIO: The New Intelligence of Organizations for the Future of Companies and Organizations.” Tranzform, 30 January 2025. https://www.tranzf.org/nio-the-new-intelligence-of-organizations-for-the-future-of-companies-and-organizations/

Morini, Maurizio. Interview by FORMart. “NIO – Nuova Intelligenza Organizzativa.” https://www.formart.it/nio-nuova-intelligenza-organizzativa-intervista-dott-maurizio-morini

Nonaka, Ikujiro, and Hirotaka Takeuchi. The Knowledge-Creating Company. New York: Oxford University Press, 1995.

Reichheld, Frederick F. “The One Number You Need to Grow.” Harvard Business Review 81, no. 12 (2003): 46–54. https://hbr.org/2003/12/the-one-number-you-need-to-grow

Senge, Peter M. The Fifth Discipline: The Art and Practice of the Learning Organization. New York: Doubleday/Currency, 1990.

Wilensky, Harold L. Organizational Intelligence: Knowledge and Policy in Government and Industry. New York: Basic Books, 1967.

About the author

Maurizio Morini is a strategic consultant and author based in Bologna. After managerial roles in industry and large-scale retail, he has worked since 2002 as an independent consultant on organization, innovation, and commercial development. He was a contract professor of business culture and related subjects, including at the University of Bologna, and directed the Fondazione Istituto Cattaneo from 2015 to 2019. He is an accredited Innovation Manager and directs the OlisTech series at Edizioni Pendragon. His books include Lezioni di Cultura d’Impresa (2004), Tecniche per un Marketing Sostenibile (2010), Zero Waste Marketing (2019), Contro la Leadership Ignorante (2019 - english version 2020), Per la leadership Evoluta ed Intelligente with Giacomo Gasperini (2020), NIO. Nuova Intelligenza Organizzativa (2024), Le Parole che fanno l’Impresa (2025), Marketing Aumentato (2026). Correspondence: the author welcomes contact regarding case permissions and independent use of the method.