This article is within the Organizational Support series.

The title of this article is a mouthful: " "Structural Knowledge and Continuous Improvement the AI Side".

Anyway, being within this section, aims to share operative tools.

And therefore, each section will share operational elements.

A small background note: in 1992, left the French company where was Head of Training and Methodologies essentially because I had a girlfriend in Germany, and planned to shift gradually there, and then we had agreed that we would actually move to Amsterdam, where we had met.

Amsterdam was chosen not for romantic reasons- but because my occupation, cultural and organization support plus business number crunching (decision support systems, financial controlling, management reporting, etc) could be done anywhere, while back then The Netherlands was the dream of anybody working in town planning.

When we met in Germany, we went around, and she shared with me all the enthusiasm about the differences and why her and others looked forward to that location.

For myself, one language or another, in a multinational environment, is "genauso"- exactly the same.

In 1991, after I had for one year declined invitations from former customers and colleagues and partners within the decision support and business number crunching domain that I had met via Andersen+Comshare (I had no formal non-competition rule, but I invented my own "12 months"), started getting again calls.

A former customer that had met while he was financial controller of a logistics company and had been a customer not of Andersen, but of Comshare (anyway, gave both the 12 months), had some activities in financial controlling where my help could be useful to build or assess models- so, checked with my CEO if adding that business would be of interest.

As that would require of course developing a structure and was outside the business, we reached an agreement: as a "cadre" after a mere few months within the company (yes, at 26 without a degree- due to business results direct and in supporting other business areas in their own development), I was told that I could set aside my overtime into a kind of "time bank", and use that responsibly, and, in exchange, if the company needed my advice on those skills that was going to keep alive and improve, I would give that advice.

As the Italian branch went into acquisition, i.e. M&A mode- and eventually we acquired a larger company and the merger was for incorporation- as laughed when I was told of the choice, the riposte was that I knew why- yes, because in that way the company had not to pay existing employees was is a kind of severance reserve that each employee accumulates, and that frankly almost no company would have available (despite being a debt against employees) if it were to be asked to release it immediately for all the employees.

It is a business reality: the TFR (as it is called in Italy) for a long time in Italy was actually used as a self-financing tool.

The arrangement was mutually beneficial: as I used to skills repeatedly also with my employer, and, on the monetary side, a single day of consulting on financial controlling etc gave me 800k LIT for 8 hours (yes, I billed the hours).

Which, market-wise, was cheap- for my own methodology and change consulting activities for my employer, billed three times as much- but, of course, there were significant structure upkeep costs, plus contribution to the overheads.

Eventually, the customer became CFO/COO of a gas distribution company belonging to a French and a Dutch group, and therefore was asked to help there.

My tradition is that things converge in a movie-style way: so, I met in Amsterdam that German woman while doing my Northern Europe tour, and then was offered to become a cadre as financial controller.

Well, as the French company, as part of the package organized with Paris to shift me from Turin to Milan and develop my business unit, dropped in also an apartment and an expenses account, it was a financial beating, also if nominally the salary was slightly more.

Anyway, the conditions were that I would have been a fractional (spending part of the week in Germany and eventually ready to move to The Netherlands, where anyway the HQ of the main owner were)- but on the day I was to sign, the conditions changed, as, again also this company went into M&A mode.

So, declined to sign but accepted to stay on as fractional management consultant focusing on business number crunching if and when needed, and started looking around.

How long did I keep supporting that customer? Until 1997, as in 1998 formally moved abroad for the first time.

A customer of my French employer, where I had developed their own cultural and organizational change introducing a methodology, discovered that I was to leave the company- I had been forbidden to tell, so I do not know who told them (probably other colleagues who feared that nobody would take care of that large customer).

So, few days before Xmas 1992, was invited to a meeting nearby Parma, where the CEO offered me a role that was a managerial role under the banking contract, but, as was needed to start right after Xmas, was to be a management consultant as program manager on that cultural and organizational change.

I had to get an authorization from my employer, but the cost of the non-competition rule was steep, so was authorized, and from January 1993 was again working on both cultural and organizational change and business number crunching.

Why this preamble? To share my "business credentials" on what you see in the title: working on both elements across multiple industries, and starting in my early 20s to work with senior management and Cxx gave me a perspective way beyond what you can get in a mere course with few fictional case studies or "sanitized" real cases.

In a real business, you see and are involved in the real impact of what you deliver, and have to cope with pre-empting, recovery, orchestration- often without having any direct official authority.

So, you have to build trust and, as my British colleagues were fond of saying in the late 1990s, "walk the extra mile": I am no alien to working overnight as a stop-gap while helping to design something that avoids an encore- not for myself, but for those who will be part of the structure.

In my management consulting activities, assessing the current status and spotting talent to help to develop (with the agreement of the management) is an invisible but "conditio sine qua non": if I am on a mission, I am on a mission- temporary, focused, and hopefully leaving behind, when it ends, something that makes sense to those left behind.

It worked before our current AI, becomes even more structural now.

Incidentally: in other articles, the AI side was confined in a specific section- in this one, as the point is, as the section title states, "organizational support", spread across.

As a formal/informal introduction, will share a short discussion about three books on "creative accounting": as it is a domain that I keep observing, and not just in balance sheets, but also within the curious way processes are executed by organizations in Italy, as saw routinely as a citizen since returning resident in 2012.

In that formal/informal introduction, without discussing about AI at all, will use three books and the rationale for reading them this summer, rationale that was inspired by initiating my full re-read of Thucydides' "Peloponnesian War"- as it resurrected so many cameos from my business experience, that decided that a read + re-read would be useful and more aligned with the potential of having to resurrect it, as in Turin and Italy keep being just treated as a Guinea pig who should be happy to work for free for those "appointed" by local tribes.

Turin and Italy since the late 1990s did just generate "value extraction"- on what I had invested or helped to invest on by others: be it nearby Parma, in London, in Paris, in German Switzerland, or elsewhere.

Anyway, this will be for a future discussion: you can read previous articles if you want to know more- but, as the local antics keeps being the same that other foreigners observed while was living in Brussels since 2005 (to force a return to Italy), I might be quite annoyed, but I think that in this article in this section anything beyond this note would be just a boring encore- so, will be just hinted in terms of issues within the "Introduction" section- in the other sections, will be discussed in terms of solutions.

I would rather share something that you can evolve or replicate, do not need moral support, as anyway, in Turin and Italy, moral support is free but generally comes with a request to deliver free advice that costs time (and often materials) to compensate for the moral one that you did not ask and was fake anyway...

The other sections of the article will be quite short- as befits something that needs to deliver "organizational support": will be developed in future material, but will share enough examples that you can actually start right now.

The second section will be on sharing a couple of results and comparisons on that "structural knowledge".

Anyway, the "continuous improvement" side has two elements: the first part will be about "cultural side"- and what it entails, the second part will be focused on number crunching, discussing something I am working on.

The table of contents:
_ Introduction: the two faces of creative accounting
_ A_Structural Knowledge
_ B_Continuous Improvement
_ Conclusions and next steps



Introduction: the two faces of creative accounting

As wrote at the beginning, this summer I had other plans.

Well, first and foremost, did not have the time and quiet I had expected- so, instead on working on completing few data projects, shifted on publishing a minibook (Citizen Audit & AI - User Manual, and the associated first episode of an entertaining series resurrecting my United Hamster Front.

Reading the full Thucydides along with re-reading Ancient Rome historians were the "reading side" I had planned.

Anyway, as one of the data projects that was planned is about number crunching Annual Reports of companies listed on the Italian stock exchange "Borsa Italiana" (see here a bit of the project history, started in 2022), reading about the wheeling and dealing of Athens and its allies resurrected memories from my past working with financial controllers and CFOs in various industries, first in the late 1980s (the most recent project in that domain was in 2024).

Hence, after starting Thucydides... did a selection of my collection on creative accounting and fraud spotting/resolution.

I selected three book not just to read, but to use to resurrect further memories, including cross-checking on organizational behavioral practices that saw in Italy on a daily basis since 2012, also when I was not working on a mission but just using the proceeds of the previous missions to dig into researching, as did since I was first registered as a freelance in 1990.

As a bookworm with a past in political activities that also in the Army found a way to set up a service, design and deliver a training, and become volunteer for many activities at the same time with the purpose of leaving the place better than I found it (which is what I assumed should be the purpose of politics), each mission or even role is an opportunity to learn, improve, and share as "assets" what you learned, positive and negative.

If you ever worked with me, you know that while I work fast and make others work fast by finding in themselves capabilities that they did not know to have, one of my obsessions is to keep track for future reuses of what was learned.

Meaning: not just the results, but also the reason why some choices where made, and what was discarded.

As the context might change, and what e.g. was not feasible when activities were manual, became feasible already in the mid-1980s, when companies started having a PC on each desktop.

So, if you had made a decision or process design in the 1960s assuming the organizational ability to process data, in the 1980s should have been revised.

And also in the late 1990s, while working in Paris, remember customers within the retail industry telling me that due to data volume they could do a detailed number crunching only at some levels.

Jump into the late 2010s, and you have vast amounts of information and cloud-based data storage.



Why those three books?

Frankly, the first one, "Accounting for Growth", is the first book on creative accounting that purchased.

The second one is about something that saw from the banking reports side across few decades (e.g. "Too Big To Fail"), but was always from the "technical" side.

Hence, was interesting reading a narrative book that was integrating the "technical" element on what organizational culture can generate, and how cognitive dissonance has real and self-referential consequences (a theme that, being from Turin and Italy, observed well enough that already in the late 1980s had had enough and accepted in 1990 to move to Milan for the French company).

The third book instead is something that covers both the first and second book, and the aftermath.

While the first one (in that edition) references cases e.g. in the 1980s and before that either observed or read through media, the second one, released first in 2009, nominally on Lehmann Brothers, covers the full industry cognitive dissonance that generated that chaos, while the third one is the 25th anniversary edition published in 2018.

So, with those three books, you can have a perception of continuity and "technicalities" about creative accounting, but also the human side.

In my experience since returning, I was amazed at how often the overall architecture of the interactions between State and private organizations in Italy is more about continuity of the existing business relationships, not producing results.

Continuous improvement is announced- but, in a country as tribal as the Wall Street financial industry described in the second book, that is to be expected.

And you can see the result... not just in debt, but in the continuous stream of announces that I saw since 2012 about "this is the solution": government after government, while our socio-economic structure deteriorated, the arrogant bureaucracies increased their own sense of entitlement and infallibility (not even the Pope has that "dogma" anymore", and each issue was just the catalyst for...

... more announces and more debt that was not an investment, but a support to continuity of what did not work.

The creative side of "creative accounting" is not about accounting- is about culture, specifically organizational culture.

This section will actually serve as a "background" to the other sections of this article.



A_Structural Knowledge

What is structural knowledge?

It is not necessarily "formal knowledge"- it is mantra in cultural and organizational change that most of the "culture" side that makes organizations work is never formalized in paper, knowledge management systems, etc.

Anyway, many mistake "structural knowledge" for formal + informal knowledge within an organization.

In reality, it is highly context-dependent.

To make a simple case: if you are used to live in an industrial country where over 70% of the people lives in urban centers, it is only natural to think that your mindset is tailored for that concept of "normal".

So, a case that shared in the past was when in Tuscany years ago there was a significant amount of snow falling in a relatively short period of time.

The Mayor of a small village said that people coming from living in major towns (Milan, Turin, Rome, Florence) were used to have everything there and e.g. having the roads cleaned in hours.

Then, they moved to villages as his own- but decided to live outside the village, to "return to the country side"- but still expected to have the same level of services.

As was told decades ago by the administrative director of an utility in Turin, after they had been required to integrate their counterparts delivering e.g. water distribution in villages outside Turin: the single village where my parents lived had less inhabitants than a single building block in Turin.

Still, expected the same level of water quality, while having kms of pipes to be maintained for almost no inhabitants.

Assessing the context requires a willingness to share assessments across the structures involved.

And requires then a further organizational willingness to distribute that assessment, so that it can be used to work on the operational culture and associated structures.

If your culture is reactive (as will be discussed in the next section), the risk is that you will even avoid assessments as a routine- you will just go for minimal compliance and continuous confirmation until a crisis requires a change.

Most crises I saw developing were actually due to a lack of structural willingness to do more that ex-post, "fixing", crisis management.

Within the European Union, in 2020 with the Recovery and Resilience Facility and its hundreds of billions of EUR became fashionable to utter the word "resilience".

Then, we created "resilience bureaucracies": instead of "active listening", "active definitions"- of resilience, of its adaptability element, etc.

If you accept that structural knowledge is both formal and informal but also highly contextualized, then probably you have to cross-check if your own knowledge management, career paths, and overall investment approach (including in unlearning, relearning, learning) are consistent with what you just accepted.

Which is why, decades ago, while working on cultural and organizational change but also when being asked to support startups, often declined the tremendous opportunity after minutes of interaction: change requires a mandate from those who can have impact on choices and "remove roadblocks".

If I get a mandate from somebody who cannot even control his own direct office, it is not a mandate- it is an invitation to act as Guinea pig crossing a minefield, so that they can see where the mines are.

Structural knowledge in our AI times generates a potential.

For example, in Italy, it is common in politics and business to be "reactive"- not to expose yourself until there is a consensus.

Then, suddenly, when somebody starts building consensus, everybody jumps on and starts explaining why they actually were heralds of that new concept- also if before wrote or communicated just the opposite.

In many organizations, a current trend is to convey the formal side of structural knowledge within an AI model, so that you can claim that also when some "key person" retired, that attitude and knowledge is maintained.

Wrong.

You got the face value in that specific context- not the attitude.

Instead, you would need the attitude as well as the "replaying" of past delivery.

In May 2026, as an experiment released on my HuggingFace profile few models that are basically composed by few elements:
_ the underlying Qwen model
_ training on the summaries and main themes of my material (for articles and for the AI ethics primer; for a dozen of my book on change, the full material)
_ some rules that try to restrict access to the material only, while using the capabilities of the underlying model to ease access.

Example: except one (in Italian), all my mini-books are in English- but you can ask questions in all the languages that Qwen "understands" (around 100).

My articles and mini-books are not about giving access to a repository of cases to replicate- but about attitude.

An Italian writer about Italian bureaucracies (mainly public, but also private) and politics is Sabino Cassese.

His books often have the same approach that I had learned as a kid by reading biographies of people who had a specific expertise.

It would be interesting, as a connecting element, to have e.g. the schools that train Italian bureaucrats (we have no real equivalent to the French ENA, as we do not have a real equivalent, in science and technology, to the German Fraunhofer), to have him prepare some "theme-focused" papers to complement his books, and then train a model on all that material.

As this would provide more than the "formal" side, but also the "contextual" and "informal", as well as the attitude- enabling a "virtual avatar of the Professor Cassese" to assist future generations.

I selected his case because he has a specific focus and high (potential) influence- not to compare myself with him, of course!

My focus is on "methods"- change, cultural and organizational change, business number crunching.

His focus and value-added is more on what instead cannot be replicated- and would be a waste to lose.

In many large organizations, what already in 2003-2005 in my e-zine on change described as "organizational memory" and how to develop it, is routinely depleted with each retirement- and replaced by layer upon layer of procedures: i.e. "frozen in time".

Anyway, once you have integrated structural knowledge, you have to anyway consider the evolving context.



B_Continuous Improvement

In the late 1980s, while working on a banking project, I remember that, as the customer was already preparing for Y2K (notably: shifting to the year having 4 digits, as otherwise after 99 for 1999, there would be 00, and many software applications would fail to work, as e.g. 99 > 00, but 1999 < 2000).

People who had not touched the keyboard in decades or had already retired were required across the whole banking industry to come back and "fix" to adapt to the new conditions.

Until recently, whenever somebody retired, it was a routine to transfer knowledge, but most of the transfer, unless there had been time to pave the way and build up gradually, was focused on the formal and, for the informal, just the most frequent- as wrote in the previous section.

I could call it the "Spartan approach to swimming": somebody retires, you can get days or weeks of "knowledge transfer", and for a while you "float" while learning.

Actually, in many "recovery" missions, I had a nominal few weeks of "transition"- that were actually few hours: so, I know what I am talking about- across multiple domains, industries, organizations, countries.

There is an element that in some countries (such as Italy) is deep within the culture: reactive more than proactive, as, was told already in the 1980s "do not approach what has not been assigned to you, because, also if you discuss or propose to fix it, then becomes your problem".

When tinkering becomes a structural element of the organizational culture, you get a continuous stream of adjustments presented as "reforms"- as it is easier to justify a "temporary" fix than to try to redesign.

This is not continuous improvement.

I worked in Italy both in companies that redesigned processes and companies that avoided doing so: and the key issue was that the different approach implied also that transferring people between the two cultures requires a degree of mandate from the top that often is lacking.

In other countries, there are similar elements: it is innate within most larger organizations and structures- but it is not a "manifest destiny".

Now, if you pile up tinkering, in the end each fix might disrupt some previous fixes, up the point when you get Gordian knots.

Hence, you develop a reactive approach: "do not fix it unless it breaks".

Trouble is, when it breaks, often nobody remembers why was done in that way- they derive a concept of the "why", but have no information about the actual trade-offs that were implied.

Even worse: often, if enough time passed by, they will use their current context as a projection of what was the original context.

It is always entertaining to attend a meeting where nobody present back then, but now nominally covering the same role, assumes what others were doing and thinking but by project her/his own assumption of what is "normal".

Continuous improvement requires being open also to the possibility to explore what eventually become blind alleys.

A current trend is about developing "world models"- e.g. physical reality models, but also models that do not need millions of books to "learn": interact to learn and build up on feed-back.

It is still to be seen how this will impact- as, whenever somebody writes "it is how we human learn"- they forget that, as somebody more relevant to our collective knowledge wrote long ago, humans a social animal (political- but of the "polis").

I could give you a serious academic reference- but instead prefer to share this one.

And, yes, while in business or also in social activities that I select might end up being team-oriented, generally I am quite happy also to have silence- and the Gossipville (Turin nickname given by somebody else over a decade ago) intensive experience since 2012 just increased that.

I will let you read the article to see the definition.

Also people like me who like to be alone anyway had a phase when they were surrounded by people: if we build models that have a "world view" but never "socialize", or never interact with "somebody" who has already been around, I wonder which monsters we will create.

Continuous improvement on the cultural side requires choices: and choices require both a context and a prioritization.

As, despite what many think, we humans do not have infinite resources- and are inclined to change our environment.

For now, the AI contribution of "traditional" (i.e. current) AI models to continuous improvement could actually be based on what I wrote in the previous section, i.e. building models that retain not just the formal (and hopefully also the informal) side of structural knowledge, but also try to model the "attitude".

In this area, probably we should shift from technologists to social scientists the model building activities- I know that sounds a bit as Asimov's Psychohistory (but also the human real side of it).

In my library, I have books from the 1950s - 1960s where was a common practice in the West (I do not know in USSR) to publish "profiles" of leaders from beyond the "Iron Curtain"- without even meeting them.

To build models able to support on continuous improvement by pre-empting potential, we need actually to add to HR functions a level of understanding (and practice) of AI technology that could be difficult to balance- and could generate some risks.

What is the ethical side of having "profilers" feed a model with a profile of a key employee, and then have said employee act to validate the model?

And where will the "ownership" reside?

If I retire from a company, and want to use my experience and personal approach to reality, should I compete with a model that my company forced me to train to "have me after they stop paying me"?



Conclusions and next steps

This is a short article, but I hope that the material above will give some ideas.

And initiate some projects- you do not really need me- you need insiders with insiders' knowledge, plus a limited and "confined" use of technologists.

I plan to further develop the ideas.

Meanwhile: have a nice week-end!