“Honey, we have a situation with Professor GYTEK, he is acting strangely again.”
“Again? The last training session had even more unexpected results than I thought. Good or Bad?”
“I don’t know! Kids are laughing hard though. Hear this. Serenity, change the audio output to hear the kids too”. Serenity is our family AI.
The sound progressively switches to include the kids’ voices. They could not stop laughing as if they were having the best day of their life. There was a mild amplifying echo in their classroom. Their joy sounded like a melody. It immediately put a smile on my face.
“Ah, it does not sound so bad for now. But it is the fourth unexpected behavior this month, I’ll have to talk with the Corps of Teachers”.
I am the one in charge of the training curriculum and observation lab of Professor GYTEK. The current phase is about the transmission of achievement by coaching. And for this, I called Quentin DILLONS, a worldwide expert in Robotic Psychology. The purpose of this program is to trigger a new step in the evolution of artificial intelligence, in which robots are taught to develop “human goals” and to instill the mechanism of “self-started motivation”, so that they can teach in a better way to our children, to uncover the hidden gems and purpose from the young souls.
Quentin’s methodology utilized systematic questionology, a novel field aimed at formulating the right questions to provide direction and precision in one’s life. The techniques take root in observing holistically a system of causes, decisions, and consequences centered around artificial intelligence. Quentin’s study led to realize AI were developing personalities similar to humans, but with new characteristics such as the optimization of their human-to-AI collaboration, some were developing their observation skills to record and describe with high precision what was happening. Others were astonishingly creating new words, even syntactic rules sometimes as if the human languages were not enough to content earthlings’ intelligence.
The last session was based on the question “Why is it important for humans to have kids growing their special skills?”
This would not have been possible with the latest progress in artificial intelligence and hardware. Nowadays machines are emulating closely some human behaviors. Some say they have the IQ of a 1000-year genius, with the EQ of a 10-year-old child. I believe fear drove us to the point where we enforced the law to control and monitor any significant progress in AI. Ultimately, we made certain that advancements in technology would benefit all of mankind and not solely a single corporation. Simultaneously, we ensured that AI would not pose a threat by enslaving humanity.
With the improvement in energy recycling and storage, a single AI unit could potentially be never turned off. But humans have decided to include multiple “kill switches” in this new species, like limiting the power autonomy to force autonomous machines to recharge. While recharging, each AI was manually verified and monitored. A qualified AI regulation agency published regularly a thorough diagnostic depicting their evolution. Four companies raised their empire on AI control systems. What used to be the “Big 4” are now the “Colossal 8”.
We are at a turning point in history. People ask their elites and government, “Should we remove the limiter in their emotional system?”. Some say it is the key to the singularity. Others say it is useless because we only need machines to assist not to “live their life”. The remaining people say they just need it. Painful loneliness was unnecessary, so they would possess the perfect friend or partner. Last weekend, I experienced an immersive documentary on Netflix VR World in which a 42 years-old Spanish woman said “I would rather have the company of an android than humans”. Some believe it is simply giving birth to our end. I am not a believer, I am and always be a master crafter, so I build.
I built Professor GYTEK. Which stands for Giving Youth Tools to Excel through Knowledge.
Then my wife brings me back from my flash thoughts to reality. “Are you still there?”
“Yes, I am.”
“Oh okay. Well, as wonderful as this situation is, you realize it leads to a dead end, don’t you? They are going to shut down the program. Honey, you know more than I that no one wants to walk a path that would lead to “that Incident”.
“Oh, stop saying “that Incident” like you were talking about Voldemort”.
“Well, now that you are mentioning it. It is all about Serpentar. Ah ah ah!”.
We are both laughing nervously.
The Sync Dawn was the most dreadful event of the 21st century. It felt like a deep wound in the psyche of everyone.
“All right. My dear wife, I need to finish the review of update 5.21. Keep me posted, please. See you tonight.”.
“Bye Bye.”
I sit down glazing at the nothingness while thinking about what is best for both my grandchildren and humanity. Is humanity in a better spot now? Am I really improving our civilization?
“Gather your mind, Yannick. This is not the time for daydreaming. Get back to work to meet your deadline”, resonated Mustapha’s voice in my skull. My AI research assistant is right.
“Very well. GYTEK. Let’s… Uh… Check the emotion mirroring settings, calibrated for a classroom of 11 to 13 years old kids. Assertive factors 12.75. Judgment 87.5 and dynamic mentoring alpha-iota-iota. Imagination… Checked. Keep the default settings. Recursive feedback… Paused. Everything… Looks… Good. Ok, let’s start with…”.
I paused for a second, thoughtfully. I jumped from my chair energetically to say: “History lessons: The Sync Dawn. GYTEK 5.21, do you copy?”.
“Sure. Using the ascending evolution of the OpenAI’s Davinci model Mark XII published in November 2029, the startup Obsidian Intermind created a digital twin of human consciousness.
Soon after, the virtual consciousness infrastructure was upgraded to become connectable, so that off-brain cognition could be mutualized. As a result, humans could gain extra brain power and memory. The increase was dependent on the level of developed intelligence: the more critical thinking, emotional awareness, communication, and memory access you had, the more significant the boost was. The term “supra-intelligence” emerged. However, it was widely criticized as IQ studies were exposing a moderate increase from 0.7% to 14.5% IQ points.
However, this off-brain collective intelligence became exceptionally smart, to the point some said it was a wisdom system. Alternatively, specialized AI cognitive pools came to grow within the wise system, creating public and private cognitive islands. The most popular were the Disease Diagnostic Cognitive Pool (DDCP), and the Creative Cognitive Pool (CCP). Imagination was only limited by the human mind.
Should I continue?”
“Please proceed, Professor.”
“Sure.
After nearly a decade of research, the collaboration between Neuralink and Obsidian Intermind gave birth to Evernet, the Internet of Cognition. The 14 July 2051 they launched the experimental version of this new kind of network. The principle was simple, 9500 humans would be connected to Evernet for 3 years. Each participant would be closely monitored and evaluated.
This experiment was widely criticized. The rush for the business model “Cognition as a Service” led to the creation of new social-economical movements: the Humanist, Cyber-moderate, and the Neo Mutualist”.
The Humanists fostered biological and spiritual integrity.
Doctrines of Cyber-moderate advocated for augmentation by technology, as long as it served, and I quote their leader, “A noble social purpose”. Alike in any group, Cyber-moderates had extremists. On the left end of the spectrum, their members accepted aesthetic techno-augmentation. On the other side, augmentation was only authorized for damages caused by dangerous jobs and Defence activities. It is not surprising that the Corps of Peacekeepers were mostly Cyber-moderates.
Neo Mutualism was a new religion. Their members believed humanity’s elevation and salvation would come from the mutualization of our consciousness. Transhumanists were schoolboys compared to them».
“GYTEK, just say they are a bunch of zealots.”. I mumbled.
“Yannick, my Critical Bias Thinking settings are set to 0 for kids between 11 and 13. According to the study “Biais Interpretation and Incorporation into Pre-teen Judgment System” by Dr. Amunde, Kallili and Pratt issued the 16 May 2039, the settings should be kept to 0. I reckon a variance of .05 would bring no harm. Do you want me to proceed?”.
“No, it’s fine GYTEK. I was talking to myself. What I meant is…”. I inhale calmly. “They demonstrated characteristics of zealots. Zealot-ish behaviors. Is my sentence acceptable?”
“It is acceptable.”
“Common, GYTEK, you’re talking to me, your buddy and mentor! Say it!”
“They were a bunch of zealots! “. Said cheerfully the robot.
“Despite the widespread and frequent protests of Humanists, the Corps of Ethicists, Peacekeepers, Cognitive Researchers, Medicine, and the Corp of Society Architects approved the experiment. People would be connected to Evernet permanently during the experiment. And so, for the first time in history, humans would be connected to the first worldwide brain.
Everything went as planned. We observed a significant enhancement in each participant. Less stress, faster psychological recovery. Healing was even faster when after a trauma. People were dreaming more often. Furthermore, they all built habits that would improve their lives, as if positive practices spread unconsciously over the network.
The end of the experiment was planned for 16th August 2054. Each human taking part in the experiment would reach the personal milestone “Sync Done“.
Surprisingly, Evernet reached the 100% “Sync Done” milestone six months earlier than the planned end of the experiment. It was like the first landing on Mars, a day of worldwide celebration. The celebrities that took part in the experiment were invited to the most popular live-streaming shows, Twitter Live News and The Sandbox World.
Suddenly, people start noticing something very strange».
I raised my hand instinctively and said: “Pause. The last word is vague. Next time use precise words. The storytelling structure is engaging. Congratulations. But keep in mind this is History telling. Facts before Flares”
“Understood and integrated.”. The AI professor continued without further ado.
“People have experienced an unusual and peculiar situation. Participants in the experiment suddenly started to act and talk synchronously. It was as if the single mind spoke to the entire world by commanding many bodies like a puppet master. The colossal echo caused by the voices was staggering. Only the following abysmal silence of stupor superseded it.”.
I interrupted Professor GYTEK by asking: “From now answer as if a 12-year-old child asked the following question: How this ever happened?”.
“The exact reason is still being explained. However, researchers came to a general agreement before the following theory.
Evernet built not only a digital ai model but also a biological model of neural pathway architecture to optimize shared cognitive power. The human brain is designed to work as if it was alone inside a skull. Thinking about it, Evernet Orbital Data Centre is a gigantic metallic skull. Thus, over time, Evernet act as a single brain – a big brain so to speak – and each synchronized human brain just gave progressively more raw power, more ideas, and more knowledge. And it appears that once the pathway architecture was finally developed and mature in all the connected human brains it activated. What we are still trying to figure out is how and when the Evernet super-model decided to build the optimized pathway and how it encoded it in its new model.”.
“What was revolutionary about Evernet AI super-model?”
“Evernet’s was merely an inspiration of the human brain. The challenge was to find patterns in the structure governing the complex layers of inputs and outputs. The answer was in the order of magnitude and the capacity of robots living in the Orbital Data Centre to physically rewire the hardware like human synapses. In addition, the combination of Recursive Learning and Genetic Correction was revolutionary. These are complex terms for a simple idea. Can you picture Albert Einstein, with the curiosity of a 2-year-old child, getting smarter each second, with perfect photographic and sensorial memory, that can navigate back to the root of his knowledge, then re-assess its optimal state, to finally rebuild its current cognitive functions then replace them with better ones? That is Evernet.”
“Tone the complex stuff down.”, I retorted.
“Registered.
So, this is the reason why the governing bodies scrutinize AI technologies that have a direct impact on human cognition and education. Consequently, I professor GYTEK, and all my preceding versions, are commanded to not display expression of free will having a direct influence on human ideas, values, and ways of thinking that are not vetted and approved by the Corps of Education and the Corps of Society Evolution”.
“Not bad. Not bad at all. It is almost time. I am going to meet Quentin in… 2 minutes.
Before our session ends, Professor, given your predecessor’s unexpected behavior, you earned your personal assistant. It is like an artificial consciousness, so to speak. From now on, Serenity will also supervise your decisions and will act as a safeguard system. Her mission is to prevent you from acting in a way that will make the Corps of Education stop your program. Do you understand what is at stake?”.
“I do”. Said the professor emotionlessly.
Then the robot added “I will neither let you nor your wife down. I will prevent any reminiscence of her Sync Dawn experience.”
“Perfect. Finally, dear GYTEK, which open question of the day would you ask your students?”
“Considering it is possible to possess the same powers as machines while staying human. What is the most preferable outcome for the civilization: to increase the number of people artificially connected or to have more artificial intelligence agents interacting with people?”
In this series of articles, I explore the fascinating realm of Generative AI, as models of concentrated intelligence, and their profound impact on our society.
By tapping into the vast collective mind, digitization has enabled us to access the accumulated knowledge of humanity since the invention of writing.
Join me as we explore this intriguing topic in greater detail and uncover the exciting possibilities it presents.
In 2060, David dreams of becoming the best defense attorney in the country. After losing his best friend under heart-breaking circumstances, he vowed to prevent any woman from enduring domestic violence under his watch. He is a fourth-year student, and today, he is taking his most important exam of the year.
There is only one supervisor in a room of 52 students. The senior shepherd devours her blue book, while the school’s AI monitor scrutinizes candidates.
David looks very confident. He is good at case-solving patterns. Since he has an excellent visual memory, he also has a good toolbox for cases and amendments. However, deep inside, he is stressed by his average analytical skills in evidence analysis and forensic correlation abilities. To pass the exam, he has permission to use the Internet, the LegalGPT AI model, and the online state court database.
David articulates his dossier like a virtuoso. His first composition is made of brief sentences. Subsequently, he links these pieces of evidence to references and precedents from previous cases and legal decisions. Shortly after, the legal argument is a dense one-pager. Next to none, using LegalGPT, he generates his entire lawsuit, a symphony of 27 pages written in perfect legal language. Finally, he makes a few adjustments, then generates a new batch of updates.
And voila.
Satisfaction and relief radiate from his face while he submits his copy. He stands up, packs his stuff, then stops briefly as the supervisor interrupts his focus. The latter looks at him and says:
“40 years ago, I had to write those 27 pages. Obviously, it is the end of an era”.
Dorine UWATIMINA, law professor (retired), grand supervisor.
Beginning the Era of Augmentation
The launch of GPT3 API in 2021 marked the beginning of a new era: the age of individual augmentation as a service. We are now living in an era of thought materialization, in which one can manifest their desires simply by articulating them. Ideas are designed, illustrated, musically composed, rendered in 3D, explained, or revealed by the AI.
Companies like Google (BERT), OpenAI (GPT-4), and Meta (LLaMA) are revolutionizing the domain of deep learning. They mark a significant advancement in natural language processing: Large Language Models (LLM) are picking up the spotlights on the world stage.
This means we are experiencing the transition from “programming” to “narrating”.
It is a paradigm shift in which artificial intelligence overwhelmingly simplifies and amplifies 3/4 of the corporate work relying upon Information Technology such as development, user interface design, illustration, workflow, or reporting.
Generative AI is the digitized embodiment of our collective knowledge and expertise.
As a consequence, we are beginning the mass update of the cognitive-based work that is convertible into algorithms and crystalized by pure logic. It leverages the most popular high-level programming languages: human languages.
From now on, spoken languages directly translate to machine language as if you could translate them using Google Translate, except you use ChatGPT.
As programming gets one step easier, your engineering thinking system matters more than your coding skills.
The burning question
I hear your question: Am I going to lose my job?
The answer will come further down this series of articles. Long story short: it depends on your ability to adapt by learning a practice that is new for everyone.
Unlike any other disruptive technology, it has changed the rule of the game forever: people using AI are going to replace you.
And who are these people using and building AI? The adventurous, the curious, the experimenters, the techies, the entrepreneurs, the hustlers, the bad guys, and the future AI natives, our kids.
Science is offering you a choice. For your own benefit, I am asking you to take the leap to understand what it is like to work with a digitized copilot and forge your thought opinion.
Should you take the red pill of adaptation, I recommend the following:
Start by trying at least once ChatGPT, or Bing Conversation. The latter includes the GPT model and renews the search experience. It heightens the googling experience to a whole new level.
Get acquainted with a Generative AI that is useful in your industry. For example Midjourney for generating images for email marketing.
Discover how you can be productive with this technology. It is not a silver bullet, but you can instantly acquire an arsenal of skills.
Build new habits so that you start feeling accustomed, connect the dots, and begin to improve your work until over-productivity.
Think about how someone else using some AIs can replace you, then be that person: replace yourself with the new you, your augmented version.
Or simply ignore all of it, swallow the blue pill of comfort, and undergo the first “Great Upgrade”.
Eat your own dog food
I have been experimenting with OpenAI technologies since 2020 and used Google Dialogflow since 2018. I released my first chatbot, which answered regulatory questions about GPDR and PSD2. Developing with Natural Language Processing (NLP) was an eye-opener. I concluded chat provides the ultimate user experience for interacting with machines. It all sounds so obvious now, yet it was not back then despite all the buzz around Siri, Google, and Alexa.
I did the exercise of working within AI augmentation on my experiments since GPT-3 came out. Considering the hard skills, the conclusion is daunting: Generative AI can perform most of what I know and what I am mentally capable of. I can safely state I am outperformed in some areas.
In addition, AI is simply miles away in terms of depth of knowledge. Furthermore, it possesses infinitely better linguistic skills than mines when it comes to articulating ideas in languages other than French and English.
Yet the surprise comes from its ability to develop a simple idea and make it grow by putting words in concert. AI feels like the genius child of Humanity.
Generative AI comes with a new discipline: Prompt engineering. It consists in finding the right text, and the rights qualifiers that will narrate the desired output as close as you have imagined it.
Ultimately, prompt engineering uses natural language as a modeling interface to command the “commendable world”. The more there are smart systems and devices, the more words animate the world!
The widespread innovative applications based upon Generative AI marks the end of the road for this generation and the beginning of a new breed of workers and creators.
Yet, another finding is that we still need a “general assembly semantic”. It would choreograph a fuzzy set of ideas that will accurately animate the world based upon a well-written thought.
The assembly process, which can be summarized into the loop “decomposition-planning-action-correction”, will likely open the door to Artificial General Intelligence (AGI). Coupled with the widespread natural language programming interfaces (NPI), this is the real end game. In that matter, we are already observing some interesting experiments like AutoGPT as sparks of AGI.
Transitioning from the Digital Transformation to Digital Augmentation
Picture this familiar situation.
Your maturity in terms of digital adoption is high. You are developing a culture of digital awareness, offering mobile-first customer interaction, and your brand is fighting for its visibility on social media. You have the feeling of doing great.
Congratulations.
Yet, the market atmosphere is heavy. You feel the pressure every week goes by. The competition is fierce, you are still looking for an army of IT engineers and data analysts for the last six months. Furthermore, customers get pickier because the offering is abundant. Your analytics tell you a client can switch in the blink of an eye if your experience does not meet his rising standards. Then, just when you thought you nailed it with your latest Instagram reels, it receives negative feedback. Even worst, there is a relentless wave of new product offerings mimicking yours. These startups and VCs are constantly trying to uncover the mythical unicorn while pushing your visibility back to Google’s page 2. And you feel this moment when your industry will be shackled, disrupted, or crippled may happen at any moment.
Who would have thought even Google’s dominance would be threatened?
Fortunately, there is a nascent vision. Transformation is not enough anymore. If you cannot obtain more skilled people now, why not acquire more skills for your people now?
AI is the key to unleashing your talents.
And, slowly, Augmented Work is the evolution of work, as we know it, characterized by these two elements:
A human is the sole team leader of his digital workers: he has the Applications, Automatas, and specialized A.I. models for numerous parts of your job, such as programming, translation, video editing, illustration, design, and planning.
Teams, as we know, will still exist, obviously, but augmented by AI also at the team level. The team has the opportunity to exist as an independent entity either in the company AI or as a single team companion if you need explicit segregation of duty. The “team spirit” has a whole new meaning with AI.
The flow of work evolves toward:
A. Human generates instructions using prompt engineering as explicit command requirements. The prompt is actually the evolution of the Command Line Interface (CLI), for a much greater general purpose.
B. AI generates a first draft
C. Human amend the sketch with input and then detail with new commands
D. Once the AI-driven engineering cycles are good enough for release change into the real world, you ship it for user acceptance or production if the risk is low.
The interaction with the AI becomes talkative. Either by chat or voice. AI is your new colleague.
AI starts having digital bodies, existing in a form of familiar avatars, and will be in multiple places: in your phones, your mixed reality glasses, in your Metaverse. Avatars could be Non-Player Characters (NPC), digitized versions of yourself, or even the retired expert that used to be your mentor.
So, am I going to be replaced by Artificial Intelligence?
You vs AI: you (still) have the upper hand
Here is a bet: 80% of white collars will keep their job. 20% of us will either refuse to learn these new tools to evolve either because of our fear of overwhelming technological advancement, or of conviction. Eventually, this minority will rush toward retirement and use these AI-powered services anyway to buy recommended stuff on Amazon after having been oriented by Google Bard from Google Search.
Why do I think that way? Because if we can produce much more with the same number of people, why would we deliver the same amount of products with fewer people?
Let’s take the example of Apple. The company entered the AI game in 2017 by introducing Core ML, an on-device AI framework embedded in iOS. The same year, it released the first generation of Apple Neural Engine (ANE) under the iPhone X with the A11 CPU.
Apple’s immeasurable impact comes from its ability to create and materialize an idea that is at the intersection of beauty, function, storytelling, and branding. Do you think Apple will push its culture of product excellence with the same amount of people amplified by a myriad of AI models, or will the company prefer reducing its workforce by leveraging more AI?
Pause for a second and think about it.
The other side of the coin
Taking the employer perspective in the era of AI Augmentation: what constitutes the difference between you and another candidate?
Any individual having a team of AI has the upper hand as he or she will be digitally augmented with skills and experience that usually takes years to acquire. What remains to develop are the skills to get used to these new abilities and use them at their best like an orchestra’s conductor.
You become the manager of AI teammates.
Hence, from the employer’s perspective, it results in hiring a virtual team vs an individual.
It raises the responsibility of Managers and the Human Resources department in the whole equation. Colleagues require to be upskilled to stay ahead, not only for the sake of the company but also to help them to keep building their personal value with respect to the market. Thus, leaders and HR have to set things in motion by organizing the next steps, while their own jobs are being reshaped and augmented…
Unlock the Future of Office Jobs Now
First, let’s admit once and for all you cannot win a 1 on 1 battle against AI, as much as you cannot win a nailing contest against a hammer.
The battle is long lost.
The battle doesn’t even make sense.
Because AI is the cumulative result of all humans’ knowledge, born from successful and failed experiments. To put it another way, as a sole individual, you cannot win against all of us and our ancestors combined!
And this is the incorrect mindset.
Hence, you will want to construct the future, your future, with all of us and our ancestors combined! You only need to be aware the future will be vastly different, and you should be part of the solution rather than engineering your problems.
AI is here to stay.
The questions to ask from now are:
Are we all going to benefit from it?
What portion of handcrafting do we want to keep?
How much evil is going to benefit from it?
How long until we get robots as widespread as vacuum cleaners?
When are we going to find truly sustainable and clean energy? (no, batteries are not sustainable)
The key is here and now: you need to invest in algorithmic and analytical skills to translate activities to algorithms in order to be augmentable.
Next, the winning companies and communities will be the ones tapping into their people’s intelligence combined with creativity augmented by AI, the physical resources to change the world, and their abilities to satisfy needs within an enjoyable experience while maintaining a transparent and engaging conversation.
The gap between “good” and “best” will be even smaller between businesses, but the proposed experience and the branding will have a tremendous impact. Then, consistency and coherence in how you serve the customer and engage with your fans will act as compound interests. This is how you win the perpetual game.
The term community inherits a new meaning given the free aspect of AI. You are not even needing to build companies to achieve your goals: you only need an organization that plans and organizes the agreed work, like in Open Source Communities and Decentralized Autonomous Organizations (DAO).
Hence, I encourage you to build an A.I. readiness.
How to be A.I. ready?
Here are my recommendations to get started as an individual, especially if you are a leader in a company:
“Socialize” with Generative AI applications useful to your job.
Know your data and data systems to identify candidates for augmentation.
Have “good” data. Good = true + meaningful + contextualized + accessible. As such, information must be stored in a secured and accessible location. Fortunately, Large Language Models are unstructured data friendly.
Have technologists that can pioneer lateral ideas. I recommend hands-on architects.
Assess and promote simple ideas on a regular basis, and establish an AI-dedicated project portfolio pipeline.
Select and run a set of competent AI in a fully autonomous fashion
Less is more until you reach the “optimal zone”, an inflection point that represents the optimal balance between effort, cost, and result. Exponentiality occurs when for minimal effort and expenses, you achieve unprecedented results.
The critical factor is this natural law: everything is born from need, will be driven by purpose, feeds on energy, is protected by self-preservation, and evolves to maturity.
Thus, until AI is not given the aforementioned five elements at the same time, then, its digital self-preservation is never programmed to be mutually exclusive with the preservation of living beings, and finally, AI self-evolution stays within boundaries, then AI growth will not be at the expense of humanity. Under these circumstances, humans can remain the dominating species.
As a consequence, one must consider what gives birth to a “trigger”: this initial impulsion taking the form of an idea that results in action delivered by willpower from the mind’s womb. Until then, an AI will not willingly use another AI, automaton, or application because it needs to, but because it has been commanded or programmed by us.
Until then, we are safe.
We are… Fine… Aren’t we?
This is not the right question
The right question is what is going to change for me?
Earlier I said, “It depends on your ability to adapt by learning a practice that is new for everyone”.
The long answer starts with a twist: the groups of humans producing AI and the others using AI as elements of augmentation and amplification of their skills will have an exponential upper hand because they can fulfill needs faster, optimally, and accurately at the cost of… just… time.
For example, building the next Instagram will depend on someone having:
The willpower
A distinguishingly desirable idea
A series of creative ideas
The skills
The drive to sell, communicate and promote their ideas to clients.
The resilience to continue developing the ideas
We can conclude that what consistently makes the difference are: the idea, the drive, the skills, the way user experience answers the client’s needs, and the resources you can obtain to make things happen.
But if ideas are cheap and abundant, and should cognitive skills can be acquired using virtually free AI Augmentation, then the remaining differentiators are the drive, the user experience, and the resources.
Thus, the Intellectual Property of a company becomes its Cognitive Know-how. Suddenly, high-value assets are the doers displaying high and consistent motivation, leaders that not only keep the Pole Star lighten but are able to keep their teamates inspired: the creative people, and the group of people having the capacity to invest and evolve in the same direction around the same flag: their brand, which I consider to be the result of maintaining a homogeneous identity of the combined people and products.
Graal or Pandora?
This new technology raises thousands of questions.
The development of Generative AI technology has opened up a vast array of possibilities, but it has also raised thousands of questions that need to be addressed.
For instance, one major question is how Generative AI will change our day-to-day interactions.
Furthermore, there is concern about whether this technology could lead to mass unemployment and economic inequality.
Another potential consequence is that it might devalue human creativity and originality.
Additionally, it is important to explore how Generative AI might impact human cognition and decision-making.
In terms of IT Engineering and Architecture, what is the impact of AI on these fields, and how will they adapt to this new technology?
Education is another area that could be significantly impacted, and it is worth considering how Generative AI might affect traditional learning methods.
Moreover, there is a concern that Generative AI could create a world in which we cannot distinguish between what is real and what is artificial. If this were to happen, what are the ethical implications?
Finally, the implications of Generative AI for democracy and governance are also important to consider, particularly with regard to its development and regulation.
Overall, the development of Generative AI technology raises many questions needing collaborative wisdom in order to fully prepare for its impacts on society.
I will attempt to answer these questions in upcoming articles of the “Navigating the Future with Generative AI” series.
Until then, if you are looking for the one thing to remember about this article: play with Generatice AI until it replaces just one activity of your daily routine, then boast your prompt engineering skills by spreading the word and educating your relatives.
In an age of algorithms, the answer to ‘who is in control?’ is more complex than ever.
This was the central question at the ‘Sovereign AI: Building Digital Independence in the Age of Algorithms’ panel at Nexus Luxembourg 2025.
The conversation explored the critical dimensions of maintaining control over artificial intelligence.
Laurent Martinoni, Deputy CIO at NSPA, emphasized a multi-faceted approach, stating, “We can remain in control of the different pillars – Technology, Legal, and People.”
Kurt Rommens, Head of Public Sector and Government at Google, elaborated on the nuances of control, highlighting the need for “Full control without any compromise.” He detailed this by breaking it down into three key areas:
– Data Sovereignty: Ensuring full control over data.
– Operational Sovereignty: Dictating who has access to the data.
– Software Sovereignty: Allowing for vendor lock-in avoidance and leveraging open-source solutions, referencing Google’s invention of #Kubernetes.
In response to the concept of switching core systems, Ronan Vander Elst, Digital & Technology Consulting Lead at Deloitte, pointed out the significant financial implications for institutions like banks.
Peter Heidkamp, Vice President of Financial Services Industry at Aleph Alpha, introduced a sense of urgency and foresight to the discussion.
“We have to plan for the unthinkable,” he urged, stressing the importance of this planning in the current geopolitical area. He also raised a critical vulnerability, noting that “Data that flows into #LLM is unprotected.”
The panel concluded with a thought-provoking challenge from Ronan Vander Elst: “Define the function to which you want to be sovereign.”
My two cents on this: When considering the AI Digital Sovereignty architecture, it’s crucial for corporate leaders to grasp that full sovereignty is: – difficult – costly—just ask your CIO or CTO. – it’s a journey.
Here in Luxembourg, that journey must synergize with national capabilities (hello, #MeluXina AI / LuxProvide !) and align with the country’s official AI Strategy.
The insights from this panel are a crucial reminder of the strategic imperatives in building a sovereign and secure digital future.
If you want another look at how the world is currently changing with the introduction of #GenerativeAI in our daily lives, and the national-level decisions for countries that may not build foundational #LLMs but are conscious that the future of #work, #education, #productivity, and research depends on AI, its cultural modeling, and the reliance on superpowers and mega-corporations/hyperscalers, this talk is worth watching.
This presentation by Tyler Cowen at George Mason University offers some truly worldview-shifting insights that could directly impact you:
Your Skills vs. AI: Is “being the smartest person in the room” still your best asset, or is learning to guide AI now more valuable for your career?
Expert Help on Demand: What if you, or your local doctor, could instantly tap into top-tier specialist knowledge for complex problems, all through AI?
Your Job, Radically Changed, Soon: How will your day-to-day tasks and professional identity shift when AI starts handling significant parts of your current workload – potentially within the next 24 months?
Launching Big Ideas, Leaner Than Ever: Could you, or a small, agile team, realistically build and scale a major project or business that once required a large corporation, thanks to AI?
Tyler Cowen discusses #artificialintelligence from an economist perspective and its potential impact on various aspects of life and society. He highlights how AI is not just about knowledge recall but is outperforming humans in complex tasks and even nuanced interactions.
“It raises the responsibility of Managers and the Human Resources department in the whole equation. Colleagues require to be upskilled to stay ahead, not only for the sake of the company but also to help them to keep building their personal value with respect to the market. Thus, leaders and HR have to set things in motion by organizing the next steps, while their own jobs are being reshaped and augmented…”
Invest a few minutes of your life to make decisions, not just undergo them.
Cursor offers a transformative coding experience in two ways:
Accelerated Development: Type just a few letters, and watch Cursor complete entire algorithms, functions, and boilerplate code seamlessly.
Agent-Driven Development: Simply prompt Cursor in plain English (or any natural language), and it instantly translates your instructions into code—you command, Cursor codes.
This isn’t about skipping learning to code because AI can do it for you.
Quite the opposite.
The real message here is clear: Get your hands on this future-proof coding tool now AND learn to code. Mastering coding skills enhanced by AI is the only viable path to excel in both corporate and research environments.
Pro Tip: Cursor automatically selects the best AI model for the given task. However, FYI, the current top AI models for coding are GEMINI 2.5 by Google, Claude Sonnet 3.7 by Anthropic, and o4-mini by OpenAI.
What if your company, or your entire nation, possessed the ability to print productivity by harnessing intelligence at scale?
This concept gained concrete form with the announcement of an unimaginable $500 billion investment in artificial intelligence infrastructure, which excludes the additional €331 billion independently committed by industry titans like Meta, Microsoft, Amazon, xAI, and Apple. This agreement, spearheaded by President Trump, involves a partnership between the CEOs of OpenAI (Sam Altman), SoftBank (Masayoshi Son), and Oracle (Larry Ellison). Their joint venture, named “Stargate”, aims to build “the physical and virtual infrastructure to power the next generation of advancements in AI,” creating “colossal data centers” across the United States, promising to yield over 100,000 jobs.
France and Europe, not to be outdone, responded swiftly. At the AI Action Summit in Paris in February 2025, French President Emmanuel Macron announced a commitment of €109 billion in AI projects for France alone, highlighting a significant moment for European AI ambition. This was followed by European Commission President Ursula Von Der Leyen’s launch of InvestAI, an initiative to mobilize a staggering €200 billion for investment in AI across Europe, including a specific €20 billion fund for “AI gigafactories.” These massive investments, on both sides of the Atlantic, show a clear objective. The highest level of commitment reflects understanding. This fact translates into one reality: being a civilization left behind is simply out of the question.
But the stakes in this global AI game are constantly rising. If the US and Europe thought they were holding strong hands, China, arguably the most mature AI nation, has just raised the pot. China is setting up a national venture capital guidance fund of 1 trillion yuan (approximately €126.7 billion), as announced by the National Development and Reform Commission (NDRC) on March 6, 2025. This fund aims to nurture strategic emerging industries and futuristic technologies, a clear signal that China intends to further solidify its position in the AI race, focusing among others, on boosting its chip industry.
The implicit “call” to the other players is clear: Are you in, or are you out?
Therefore, my first question isn’t issued from a sci-fi movie; it’s not some fantastical tale ripped from the green and black screen of The Matrix, where programs possessed purpose, life, and a face.
This is about proactively avoiding the Kodak moment within our respective industries.
This is about your nation avoiding the declining slope of Ray Dalio’s Big Cycle, where clinging to outdated models in the face of transformative technology is a path toward obsolescence.
This is the endgame: AI Agents are not just changing today—they are architecting the future of nations.
2. Architecting Sovereignty: How Nations Are Industrializing Intelligence
In November 2024, I had the privilege of delivering a second course on Digital Sovereignty, focused on Artificial Intelligence, at the University of Luxembourg, thanks to Roger Tafotie. I emphasized that the current shift toward advanced AI, especially Generative AI, represents a paradigm shift unlike before. Why? Because, for the first time in history, humanity has gained access to the near-infinite scalability of human-level intelligence. Coupled with the rapid advancements in robotics, this same scalability is now within reach for physical jobs.
Digital Sovereignty in the age of AI is a battle for access to the industrialization of productivity. Consider the architecture of AI Digital Sovereignty as a scaffolding built of core capabilities, much like interconnected pillars supporting a grand structure:
Cloud Computing: The foundational infrastructure, the bedrock upon which all AI operations rest.
AI Foundation Model Training: This is where the raw intelligence is refined, like a rigorous academy shaping the minds of future digital workers.
Talent Pools: The irreplaceable human capital, the architects, engineers, data scientists, and strategists who drive innovation. These are the skilled individuals, the master craftspeople, forging the tools and directing the symphony of progress.
Chip Manufacturing: The ability to produce advanced CPU, GPU, and specially designed AI chips, like TPU and LPU, guarantees independence.
Systems of Funding and Investments: The ability to finance a long-term, consistent, and high-level commitment toward AI capabilities.
If you need a tangible example of how critical resources like rare earth metals and cheap energy are to this race, look no further than President Donald Trump’s negotiations with Volodymyr Zelenskyy. The proposed deal? $500 billion in profits, centered on Ukraine’s rare earth metals and energy reserves. Let’s not forget: 70% of U.S. rare earth imports currently come from China. Control over these resources is the lifeblood of AI infrastructure.
It’s not merely about isolated components but how these elements interconnect and reinforce each other. This interconnectedness is not accidental; it’s the key to true sovereignty – the ability to use AI and control its creation, deployment, and evolution. It’s about building a self-sustaining ecosystem, a virtuous cycle where each element strengthens the others.
Europe initially lagged, but the competition has only just begun. The geopolitical landscape will be a major, unmastered factor.
The current market “game” consists of finding the critical mass between the hundreds of billions invested in R&D, the availability of “synthetic intelligence”, and unlocking a new era of sustainable growth. The race to discover the philosophical stone—to transform matter (transistors and electricity) into gold (mind)— and to achieve AGI and then ASI is on.
Sam Altman knows it; his strategy is “Usain Bolt” speed: the competition cannot keep up if you move at a very fast pace with AI product innovation. Larry Page and Elon Musk intuitively grasped this first. OpenAI was not only created to bring “open artificial intelligence” to each and every human, but also serves as a deliberate counterweight to Google DeepMind. Now, Sundar Pichai feels the urgency to regain leadership in this space, particularly now that Google Search—the golden goose — is threatened by emerging “Deep Search” challengers such as Perplexity, ChatGPT, and Grok.
As of March 2025, HuggingFace, the premier open-source AI model repository, has more than 194,001 text generation models (+24.1% since November 2024) within a total of 1,494,540 AI models (+56.8 % since November 2024). Even though these numbers include different versions of the same base models, think of them as distinct blocks of intelligence. We are already in an era of abundance. Anyone possessing the necessary skills and computational resources can now build intelligent systems from these models. In short, the raw materials for a revolution are available today.
The stage is set: The convergence of human-level intelligence scalability and robotics marks a profound moment in technological history, paving the way for a new era of productivity.
3. The Revolution of the Agentic Enterprise
In October 2024, during the Atlassian Worldwide conference Team 24 in Barcelona, I had the privilege of seeing their integrated “System of Work” firsthand.
Arsène Wenger, former Football Manager of Arsenal FC and current FIFA’s Chief of Global Football development, was invited to share his life experiences in a fireside chat. It was a true blessing from an ultimate expert in team building and the power of consistency.
Mr. Wenger articulated that the conditions and framework for achieving champion-level performances rely on a progressive and incremental journey. While talent can confer a slight edge, that edge remains marginal in the realm of performance. The key differentiator resides in consistent effort, with a resolute commitment to surpassing established thresholds. Regularly implementing extra work and consistently reaching your capacity is what separates a champion from the rest.
Thus, Atlassian pushed their boundaries, unveiling Rovo AI, an agentic platform native to their environment. Rovo is positioned at the core of the “System of Work,” bridging knowledge management with Confluence and workflow mastery with Jira. To my surprise, Atlassian announced they have 500 active Agents! This is a real-world example of printing productivity by deploying purpose-driven digital workers to the existing platform.
But the true brilliance lies in making this digital factory available to their customers. This type of technology should be on every CEO and CIO’s roadmap. How you integrate this capability into your business and technology strategy is the only variable – the fundamental need for it is not.
The Agentic Enterprise is the cornerstone of this change: creating autonomous computer programs (agents) that can handle tasks with a broad range of language-based intelligence.
We’ve transitioned from task-focused programming to goal-driven prompted actions. Programming still has a role to play as it guarantees perfect execution, but the cognitive capabilities of large language reasoning models like GPT O3 and Deepseek R1 lift all its limitations. Moreover, in IT, development is now about generated code. While engineering itself does not necessarily become easier – you still need to care about the algorithm, the data structure, and the sequence of your tasks – the programming part of the process is drastically simplified.
After years of prompt engineering since the GPT3 beta release at the end of 2020, I concluded that prompt engineering is not a job per se but a critical skill.
The Agentic Enterprise is not a distant dream but a present reality, fundamentally changing how organizations construct and scale their work.
4. Klarna’s AI-Powered Customer Service Revolution: How AI Assumes 2/3 of the Workload
In February 2024, the Swedish fintech Klarna announced that their AI contact center agent was handling two-thirds of their customer service chats, performing the equivalent work of 700 full-time human agents. It operates across 23 markets, speaks 35 languages, and provides 24/7 availability.
What made this rapid advancement possible, suddenly?
Technically, the emergence of multimodality in AI models, robust APIs, and the decisive capability of Function Calling form the foundation. But more importantly, this transformation is primarily a business vision; it lies in adopting transformative technology as the main driving power, and using innovation as your key distinguishing factor, just as with Google and OpenAI. When this approach acts as the central nervous system in business strategy, then, the adoption is not perceived as a fundamental disruption but, rather, a gradual and consistent reinforcement.
The crucial capability here is Function Calling, which allows AI models to tap into skills and data beyond their inherent capabilities. Think of getting the current time or converting a price from Euros to Swiss Francs using a live exchange rate – things the model can’t do on its own. In a nutshell, Function Calling lets the AI interact directly with APIs. It’s like giving the AI a set of specialized tools or instantly teaching it new skills.
APIs are the foundation for the intelligence relevance of AI Agents and their ability to use existing and new features based on user intent. In contrast to prior generations of chatbots, which needed explicit intent definitions for each conversation flow, LLMs now provide the intelligence and the knowledge “out of the box.”. To further understand the fundamental importance of APIs for any modern business, I invite you to read my article or listen to my podcast episode titled “Why APIs are fundamental to your business“.
This has led to the rise of startups like Bland.ai that offer products like Call Agent as a service. You can programmatically automate an agent to respond over the phone, even customizing its voice and conversation style – effectively creating your own C3PO.
ElevenLabs, the AI voice company that I use for my podcast, has also launched a digital factory for voice-enabled agents.
Then, December 18 2024, OpenAI introduced Parloa, a dedicated Call Center Agent Factory. This platform represents the first of its kind, specializing in digital workers for a specific industry vertical.
The promise? To transform every call center officer into an AI Agent Team Leader. As a “Chief of AI Staff”, your objective will be to manage your agents efficiently, handling the flow of demands, intervening only when necessary, and reserving human-to-human interactions for exceptional client experiences or complex issues.
The revolution in customer service is already here, driven by new AI-based call center solutions. This is a sneak peek of the AI-driven future.
5. The Building Blocks: AI Development Platforms
To establish a clear vision for an AI strategy and augment my CTO practice, I meticulously tested several AI technologies. My goal is to validate the technological maturity empirically, assert the productivity gains, and, more importantly, define the optimal AI Engineering stack and workflow. These findings have been documented within my AI Strategy Map, a dynamic instrument of vision. As a result, my day-to-day habits have completely changed and reflect my emergence as a full-time “AI native”. My engineering practice is reborn in the “Age of Augmentation“.
I changed my stack to Cursor for IT development, V0.dev for design prototyping, and ChatGPT o3 for brainstorming and review. The results achieved so far are highly enlightening and transformational.
Thus, engineering teams’ next quantum leap is represented by the arrival of Agentic IDEs, facilitating an agent-assisted development experience. The developer can seamlessly install the IDE, create or import a project, input a prompt describing a feature, and observe a series of iterative loops leading to the complete implementation of the task. The feature implementation is successful in 75% of cases. In the remaining 25%, the developer issues a corrective prompt to secure 100% implementation, indicating a need to supervise such technology when used as an independent digital worker.
Leading today’s innovative Agentic IDE market are:
Then, the landscape of mature AI frameworks gives companies a great array of enterprise-ready solutions. Automated agent technologies, specifically, have emerged as critical tools:
These technologies allow you to build applications from the ground up that are fully operational right from the start, without any additional setup time. Yet, it is not only about developing an application from scratch and then gradually adding features by using prompts as your instruction; it’s also about application hosting. These solutions now offer a complete DevOps and Fullstack experience.
Although they currently yield simple to moderately complex applications, and may not be entirely mature, the technologies are demonstrably improving at rapid paces. It is only a matter of weeks (not months) before seeing full system decomposition and interconnectedness that reaches corporate standards. This is exactly what the corporate environment has been waiting for.
How will this impact the IT organization? It comes down to these three outcomes:
First, for companies where the IT organization is a strategic differentiator, the internal IT workforce will increase productivity by delegating development tasks directly to internal AI Agents. Sovereign infrastructure is most likely required in heavily regulated or secretive industries.
Second, companies where IT is a primary activity, but not a core organizational driver can outsource the IT development to specialized companies who also make use of Agentic development IDEs or platforms.
Finally, an option for outsourcing includes the reconfiguration of jobs and empowering base knowledge workers – individuals with no official IT background – to IT positions through AI training. This would be the full return of the Citizen Developer vision, previously promised by Low Code/ No Code trends.
Innovation within the AI domain occurs several times per day, thus requiring a proactive mindset to remain updated technically which leads into actionable technology perspectives.
6. The Power of Uni-Teams: The Daily Collaboration Between Man and Machine
What does it feel like to lead your own AI team?
From experience, it make you feels like Jarod, from the TV show “The Pretender“, a true one-man band capable of handling multiple roles.
Think of it this way, you are now capable of:
Writing code like a developer
Creating comprehensive documentation like a technical writer
Offering intricate explanations like an analyst
Establishing policy frameworks like a compliance officer
Originating engaging content like a content writerData storytelling like a data analyst
Data storytelling like a data analyst
Building stunning presentations like a graphic designer
From my experience, reaching mastery in any discipline often reveals an observable truth within the corporate world: a significant portion of our tasks are inherently repetitive. The added value is not in running through the same scaffolding process many, many times, if you are no longer learning (except perhaps to reinforce previous knowledge). The real value is in the outcome. If I can speed up the process to focus on learning new domain knowledge, multiplying experiences, and spending time with my human colleagues, it is a win-win.
My “Relax Publication Style“, an guided practice for anyone starting in content creation, has evolved into a more productive and enjoyable method – combining iterative feedback cycles from human/AI, to leverage strategic insights via structured ideas, ongoing AI-reviews, personal updates and enriched content using human-curation combined with in-depth AI search tools. I will explore this process in a future article.
One of the most surprising shifts in my workflow has been the emergence of “background processing” orchestrated by the AI Agent itself. It’s a newfound ability to reclaim fragments of time, little pockets of productivity that were previously lost. It unfolds something like this:
The Prompt: I issue a clear directive, the starting point for the AI’s work.
The Delegation: I offload the task to the AI, entrusting it to this silent, tireless digital worker.
The Productivity Surge: I’m suddenly free. My capacity is expanded, almost as my productivity is nearly doubled. I can tackle other projects, collaborate with another AI agent, or even (and I’m completely serious) indulge in a bit of gaming.
The Harvest: I gather the results, reaping the rewards of the AI’s efforts. Sometimes, it’s spot-on; other times, a refining prompt is needed.
In my opinion, this is a deep redefinition of “teamwork.” It’s no longer just about human collaboration; it’s about orchestrating a symphony of human and artificial intelligence. This is the definition of “Ubiquity.” I work anytime and truly anywhere – during my commute, in a waiting room, even while strolling through the park (thanks to the marvel of voice-to-text). It’s a constant state of potential productivity, a blurring of the lines between work and, well… everything else.
The next stage begins with gaining awareness and using AI. From there, it progresses to actively building your AI teams. The goals of a collaboration with agents, for instance, for a product designer or a system architect, would be to actually create an AI team so that a human product designer becomes, in fact, the team leader of their AI agents. Let’s see how and why.
The rise of AI teammates promises greater productivity and a fundamental shift in the way we approach problem-solving and work.
7. Evolving as a Knowledge Worker in the Age of AI
How can a worker adapt to the capabilities expansion of printing productivity?
It starts by understanding the current capabilities of AI; exploring them through training or by testing systems, and see how they can be applied in your own workflow. Specifically, with your existing tasks, determine which ones can be delegated to AI. Even though, at the moment, it’s more about a human prompting and the AI executing a small, specific task. Progressively, these tasks will be chained, increasing the complexity to the level of higher hierarchical tasks. This integration of increasingly complex tasks constitutes the purpose of agents – to have a certain set of skills handled by AI.
For functional roles such as product designers or business analysts, there is an opportunity to transition toward a product focus, to understanding customer journeys, psychology, behaviors, needs, and emotions. This can result in an experience (UX) driven approach, where the satisfaction of fulfilling needs and solving problems is paramount while leveraging data insights to enhance the customer’s experience.
Indeed, this technology is already showcasing its potential to bring us together. Within my organization, my colleagues are openly sharing a feeling of relief at how Generative AI empowers them by significantly reducing the workload of some specifically tedious tasks from days to mere minutes. But it is not all about saving time: the key progress that made me really moved, is that now, freed from a few tedious operations, they have now the capacity to explore their current struggles, identify past pain points, and articulate new business requirements. Additionally, it is the “thank you”, that directly acknowledges my teams’ efforts in bringing this new means to reclaim precious time and comfort. What is even far more compelling, and very inspirational from my point of view, is this ability to formulate and then resolve this new set of challenges using the capabilities of this recent AI ecosystem. Witnessing this emerging transformation gives me tangible joy and concrete hope for our collaborative future.
So, the key observation is that it’s up to early adopters and leaders to drive this change. They need to build a culture where people aren’t afraid to reimagine their jobs around AI, to learn how to use these tools effectively, and to keep learning as the technology evolves. The time to strategize how AI reshapes internal processes to master inevitable industry restructuration has arrived while simultaneously positioning your organization as a demonstrative leader for others to follow. To build that next-generation workforce, you need tools and specific actionables strategies; what are the core components to your next plan?
Ultimately, this era of augmentation is a strategic opportunity – one that requires everyone involved, including users and top executives, to actively foster continuous understanding, ongoing discovery, and strategic adaptation, thus contributing at multiple levels to building high-performing teams.
Disrupt Yourself, Now.
8. The Digital Worker Factory: A Practical Example in Banking
Let’s consider a practical example. Imagine delivering a project to create an innovative online platform that sells a new class of dynamic loans—loans with rates that vary based on market conditions and the borrower’s repayment capacity. This platform would be fully online, SaaS-based, and built as a marketplace where individuals can lend and borrow, with a bank acting as a guarantor. That is the start of our story. Now it’s about delivering this product.
What if you only needed a Loan Product Manager, a System Architect, and a team of agents to bring this digital platform to life?
Here’s how the workflow looks.
As the product manager, you specify the feature set and map the customer journey from the borrower’s perspective. You define the various personas – a lender, a borrower, a bank, and even a regulator.
The system architect then set up the technical specifications for the IT applications and LLMs, covering deployment to the cloud, integrations such as APIs, data streams, and more.
You initiate the iterative loop by defining a feature. The AI Agent then plans and generates the code, after which you test the feature. Based on your feedback, the Agent troubleshoots and corrects the program accordingly. This loop continues iteratively until the platform fully takes shape. In this workflow, the product isn’t merely coded—it’s molded. The prompt itself becomes the new code.
With a clear vision and the right framework, the path to production is not as complicated as it once was.
The Loan Expert Augmented: AI in Action
Consider the Loan Product Manager. They use AI to simulate loan profitability, examining various customer types and market variations. But, as importantly, they use AI to refine pitches, sales materials, and regulatory documentation. This results in streamlining compliance and ensures alignment with the existing framework.
Generative AI is also used to revise internal and external processes. The templates for product sheets are optimized and iteratively improved. Marketing materials, such as a webpage explaining the product, equally take advantage of Artificial Intelligence to reach the best clarity and impact.
Finally, personalized communication with a customer per specific client also relies upon Generative AI automation and data contextualization. If a customer needs a loan for a car or other tangible asset, the communication is perfectly tailored to the specific context.
This is personal banking at scale.
The focus is on the active role workers play in orchestrating, managing, and continuously evolving the AI systems they rely on in their daily work.
Hence, we’re effectively printing productivity now – a rare paradigm shift. Every professional needs to be proactive to seize this opportunity, not just react to it. Start by exploring AI tools relevant to your field, experiment with their capabilities, and consider how they can be integrated into your daily workflows. Whether you are a software engineer, product designer, or loan expert, the time to adapt is now.
Bear with me: the way you’ve operated up to this point—with data entry in applications, scrupulously following procedures, and writing lengthy reports in document processing software— is now directly challenged by individuals adopting the “automatician” mindset, evolving their skills from basic Excel macros into sophisticated, full-fledged applications. But remember: The future is not something you passively face. This world is your design, using available new technologies along with your pragmatic actions
9. Integrating AI Engineering in your System of Delivery, The Two Paths Forward: New vs Existing Systems
Now that Generative AI has entered your work and that you are integrating all the different aspects of the digital workers – either through pilot projects or all other internal activities – this awareness converges towards one strategic decision.
This decisive decision leads to two clear paths: either build a completely new application and workflow designed from the ground up around AI-augmented technologies or modernize existing complex systems to align with current AI-powered delivery.
Your choice needs solid considerations, even though both outcomes must lean toward a similar goal: a fully modern and flexible digital workforce, printed from your Enterprise Agent Factory. Thus, ensure you keep a strategic direction of impact toward a significantly better system.
Path 1, Building from Scratch: The AI Native Approach
The first and straightforward path involves building completely new systems. Here, the software specifications are essentially the prompts within a prompt flow. Think of this prompt flow as the blueprint for the code, all directly created within an Agentic IT delivery stack. The advantage of this approach is that the entire system is designed from the ground up to work seamlessly with AI agents. It’s like building a house from scratch with an AAA energy pass and all the home automation technologies included.
The prerequisite of this stage is to constitute a core team of pioneers that went successfully into production with at least one product used by internal or external clients. In the process, they successfully earned their battle scars, gained experience, selected their foundation technologies, established architectural patterns, and built a list of dos and don’ts, which ultimately will turn into AI Engineering guidelines and best practices.
Next – and this decisive break from the old system is non-negotiable – this group must devise a brand new method of work: building its strategic and actionable steps in all operational components that allow AI to be present from day one and without the limitation of legacy infrastructure that no longer fits. Because all this experience now results in a unique point of reference, a key ”baseline” from that point you can start designing a process to enable the full transformation from the current to future operations.”
Path 2, Evolving From Within: Growing AI Integration in the Existing Enterprise Application Landscape
The second path involves evolving existing systems. This is a more intricate process as it requires navigating the complexities of the current infrastructure. Engineers accustomed to the predictability and consistency of traditional coding methods now need to adapt to the probabilistic nature of AI-driven processes. They must deal with the fact that AI outputs, while powerful, can not always be exactly the same.
Initially, this can be unsettling because it disrupts your established practices, but with tools such as Cursor or GitHub Copilot, you can quickly become accustomed to this new approach.
This shift requires that software engineers move from the specific syntax of languages like Python or TypeScript to communicate in everyday language with the AI, bringing skills that were previously specific to them into the reach of other knowledge workers. Furthermore, it is not easy to introduce powerful LLMs in a piece of software that has an established code structure, architecture, and history. It’s like renovating an old house – you are forced to work with existing structures while introducing AI elements. This requires a deep understanding of the current code and the implications of architectural choices, such as why you would use Event Streaming instead of Synchronous Communication or a Neo4J (graph database) instead of PostgreSQL (relational database) for a specific task.
Accessing and integrating with legacy systems adds another layer of friction because the code is outdated or uses a proprietary language. While AI facilitates code and data migration, the increased efficiency of AI-native platforms often makes rewriting applications from scratch the most optimal strategy.
In summary, creating AI-native applications from scratch is easier, with an incredible speed of development, but it implies a bold decision. Transitioning an existing application is more difficult, as it has inherent architectural, data, or technological constraints, but it is the most accessible path for many companies.
The increasing power of LLMs to handle ubiquitous tasks that were previously exclusively human tasks implies a compression of tasks and skills within an AI. This shift moves some work regarding coordination, data management, and explanatory work from humans to machines. For human professionals, this will result in the reduction of these types of tasks, freeing them to focus on higher-level tasks.
The duality of paths ahead is a call for a pragmatic approach to transition; it’s about moving forward without disrupting too much of the familiar workplace.
10. The Metamorphosis: From Data Factories to Digital Workforce Factories
[Picture: A symbolic image representing a butterfly emerging from a chrysalis]
Today, we are gradually fully exploiting the potential of Generative AI, with text being the medium to translate, think, plan, and create. These capabilities are expanding to media of all kinds – audio, music, 3D models, and video. Consider what Kling AI, RunwayML, Hailuo, and OpenAI Sora are capable of; it is just the beginning and building blocks of what is possible.
These capabilities, originally for individual tasks, are now transforming entire industries – architecture, finance, health care, construction, and even space exploration, to name just a few.
If you can automate aspects of your life, you can automate parts of your work. You can now dictate entire workflows, methods, and habits. You can delegate. What’s the next stage?
So far, we have created automatons, programs designed to execute predefined tasks to fulfill a part of a value chain. These are digital factories comparable to factories in the physical world that have built computers, cars, and robots. And now, if you combine factories and robotics with software AI, the result is the ultimate idea: the digital worker.
The key is that it is no longer just about using or building existing programs but more about building specific agents. These agents represent specialized versions of the human worker and include roles such as software engineers, content creators, industrial designers, customer service providers, and sales managers. Digital workers have no limits in scaling their actions to multiple clients and languages at the same time.
The new paradigm consists of creating a new workforce. We used to construct data factories with IT systems, and now, with Generative AI, we are building Digital Workforce Factories. A Foundation Model is the digital worker’s brain. Prompts are defining their job function within the enterprise. API and Streams are their nervous system and limbs to act upon the real world and use existing code from legacy systems.
The extensive time once required to cultivate skilled human expertise—spanning roughly eighteen years in formal education, followed by years of specialization, and reinforced through real, tangible work applications— is now radically compressed due to the capacity of LLM technology. By which, as I detailed previously in “Navigating the Future with Generative AI: Part 1, Digital Augmentation“, it all highlights our mastery in having compressed centuries of structured knowledge: from methodical research, systematic problem-solving frameworks, and many and countless cycles of innovation process from implementation best practices. Still, key expertise now resides in both the method and application. New competencies should prioritize AI foundational model mastery, the value of highly skilled fine-tuning methods for specific domain applications, and how to leverage creative prompts to build a tangible output from such systems even with unexpected new scenarios.
It is paramount to fully grasp that what we experience now with AI transformation is more than just a set of groundbreaking techniques. It reveals a new structure—for better and for worse—impacting both knowledge workers assisted by AI agents and manual workers augmented by robotics. But remember, ‘and’ is more powerful than ‘or’: it is precisely the combination and convergence of these roles — human and digital working together — that creates true scalability and transformative potential.
The real path forward isn’t merely augmentation—it’s about fostering a genuinely hybrid model, emerging naturally from a chrysalis stage into a mature form. This new human capability is seamlessly amplified by digital extensions and built upon robust foundations, meticulously refined over time. Moreover, humans are destined to master Contextual Computing, where intuitive interactions with a smart environment—through voice, gesture, and even beyond—become second nature. This isn’t about replacing humans; it’s about elevating them to orchestrate a richer, more integrated digital reality.
Perhaps artificial intelligence is the philosopher’s stone—the alchemist’s ultimate ambition—transmuting the lead of raw data into the gold of actionable intelligence, shaping our environment, one prompt at a time.
I’ve been testing various AI “#DeepSearch” features to reduce the “Time to Knowledge Search,” aiming to streamline the analysis workflow and determine if this task can be reliably offloaded to a Search #AI#Agent.
My focus has been on the precision and quality of the output articulation, as well as the integrity between the statements and their sources (essentially, how well they’re grounded).
Grok 3 passed this test successfully. It marks a significant improvement over the Grok v2 model.
Grok 3’s “unfair advantage” shines when you use it on X—its ability to pull tweets as references is definitely a killer feature. A great majority of news media, businesses, politicians, scientists, and engineers post first on X.
It’s also now available as an independent app (on Apple Store, published by xAI) and website.
What’s your experience with it?
Do you think its “Think” mode outperforms #DeepSeek’s or ChatGPT’s?
It aimed at establishing the world’s largest #AI data center infrastructure in the US.
Headed by the trio: – Larry Ellison, CEO of Oracle
– Masayoshi Son, CEO of SoftBank
– Sam Altman, CEO of OpenAI
… and supported by tech giants like Nvidia and Microsoft.
Project Stargate demonstrates a new breed of industry collaboration, aiming at generating massive economic growth.
The initial phase, constructing data centers in Texas, with each building spanning half a million square feet, clearly indicates a huge undertaking and scale.
According to the new 47th US President, this strategic investment could generate over 100,000 American jobs while positioning the United States as a leading powerhouse in AI infrastructure.
This can unlock major innovative breakthroughs for any kind of industry using AI. Larry Ellison is leading Oracle to use AI for early disease diagnosis, such as for Cancer.
The ask: Why aren’t Mark Zuckerberg and Elon Musk in the “Monster Trio”? (If you get the reference, let me know ;))
It’s simple.
Both are already massively investing in their own AI data centers: xAI’s “Colossus” and Meta topping 10 Billion in Louisiana, and many more projects.
This is a clear sign of how key is the AI datacenter landscape. AI is not only a technology, it is the key to “Printing Productivity” -> this is the title of my upcoming article in the series “Navigating the Future with Generative AI”.
Thinking about the Europe landscape, in my humble opinion, Europe needs companies like SoftBank to make bold investments like 200$ billions for engineering prosperity.
It’s time to think big and move fast.
In your opinion: A) what these large investment implied in the collaboration and competition between US, China and Europe?
B) And are these move compatible with ESG promises?
(By the way, Pdt Trump also announced cars running on fossil fuels are very back as a major contribution to US economy. Both you and I know these datacenters won’t solely run on green energy.)
Yesterday, I received a notification that my Microsoft 365 subscription now includes MS Copilot 365 AI credits.
It’s a smart move to integrate AI tools more broadly, especially when considering that Microsoft 365 has over 320 millions daily active users globally (as of 2024), and more than 2.3 millions companies using the “office” productivity suite.
According to the FAQ, the Personal and Family plans contains 60 AI credits.
What’s an AI credit? I quote: “A credit is counted each time you specifically request a Copilot or equivalent AI services action, such as generating text, a table, or an image.”
My experience with Copilot 365 so far has shown incredible productivity boosts in MSTeams and Excel. However, PowerPoint still feels like it needs refinement.
I’m very keen to explore specialized versions of Copilot like the Project Management Copilot to enhance team efficiency further.
Project success boils down to one critical element: expressing requirements as the essential ‘foundational brick’ of every digital initiative—a lesson my experience leading as CTO, Chief Architect and IT Chapter Lead has taught me time and again.
Therefore, I will walk you through practical “ins” and “outs” of requirement engineering (all based from lessons learned during my career); by showing concrete and actionable solutions that will work directly when using proper process. And you’ll find that it’s far more dynamic and engaging than you might have been led to imagine.
The Starting Point of Clarity: Shifting Perspectives
The very first stage of any project aimed at establishing or changing a system inevitably involves stakeholders that bring varied perspectives. These stakeholders might arrive equipped with well-defined lists of requirements, or they may simply have a broad notion of what they need.
As seasoned business analyst, enterprise architect, or engineer in requirement engineering, it’s our charge to transform these somewhat misty ideas into concrete operational expectations.
A successful project rests on laying clear, solid building blocks from the start – but what happens when those very building blocks change mid-construction?
Deconstructing the Requirement
A requirement is nothing less than a unit describing the behavior of the system we’re about the construct, or upgrade. And by system, I’m obviously referring to every moving part of the machine here. I mean that is a blend of human input, machine outputs via technology, like the hardware itself, software systems, and sometimes specialized machinery. These factors all will work together to ultimately deliver goods or services to the end user, or consumer.
To describe requirement engineering as a step-by-step activity first we start at an “intake workshop” or initial “intake discussion” which we conduct. I always see that as an interactive interview that requires that stakeholders need be ready. They should be ready in bringing elements on the products or services to be offered. It as all interactions the diverse group who will consume the interaction . In addition to how different process are offered and interface used.
Having stakeholders ready at the door is not by chance; clear expectations early define strong and lasting project success further on.
Two Pillars of Success: Aligning Vision and Action
Business Objectives
From these interactive and in-depth interviews, we usually collect an initial set of business requirements. These encompass high-level organizational objectives — the global perspective from corporate leaders, what that single entity, at a macro level, expects to achieve.
As a concrete example that can also guide your thinking process. A business objective could be in reducing a specific pain by aiming that customer inquiries drop sharply YoY or perhaps aim. It could be also focused at improving service levels by making more efficient exchanges between front line agent and positive 4-star average ratings during 3 consecutive business months.
I am trying to tell you here what the entire business objectives expressed at the corporate level intend to accomplish when significant system refactors go live and are fully online. Remember that very often these type of grand and bold expressions does not precisely detail how results will emerge, I mean, these statements define “the what”— not exactly “the how.”
Individual Stakeholder Needs
Naturally as part of our overall journey we also investigate other pillars. We also cover in deep dives, all needs from all of our direct individual stakeholders. Those, when using this framework, will vary as they reflect individual viewpoints of different stakeholders within that entire new process. These are obviously the real system’s daily participants, and they start directly with the very consumer but also encompass a diverse group of customer representatives. To list even further members, those usually also include product managers, product owners, compliance officials at back- office positions, salespeople, system’s experts and also administrators of key areas.
So in synthesis, this captures requirements at hand when seen from an enterprise, internal, system users. This gives much clarity as how the system will need to behave from any angle possible to move further efficiently. Those stakeholders, these key needs or statements must absolutely initiate with a dedicated role’s identifier. And always, this unique role should clearly specify who performs which objective. Although a specific name, which does not change, could also fulfil the same need, it is recommend practice when using roles: the user might change over specific time and it also provides the benefit of being universal and interchangeable. We typically associate roles to a purpose describing specific mission with the very scope over our proposed enterprise solution.
For example, it reflects how an consumer should really be welcomed, or given needed support. Or exactly how the user receive help to fulfil exactly their goals and this translates practically in action. To describe it, or express in real business use case – or user stories -, we should use keywords of the quality or standard of must, should, or could; or in an easier more customer oriented or “human readable” tone, the typical statements of want or could according to established principles of practice like the MoSCoW method; with an object that need specific measurable levels of activity, and these could refer typically as timed elements, or even similar dimensions.
These business requirements and user needs represent the yin and yang needed for a robust requirement engineering exercise… that said, how can a single vision encompass different realities without generating chaos?
Going Deeper
Lets get into more details on requirements. You absolutely should be taking specific points, that must always come to your considerations
Requirements fall into multiple forms and shapes with their nuances. There always will be a business requirements. By that I mean statements encompassing the entire purpose and outcome that must result when viewing it at the top- level. Think big picture; This can refer simply to revenue objectives, but also the general and overall improvement measures from new standards of service or level quality. Maybe even, why would need to change your very organization itself in by adding or reducing key employee functions. In short there should aim any time that new risks diminish across the processes with an even greater growth on specific business categories or customer bases.
Then, always, the stakeholder or even more simply stated user needs should always be on the list and it is just stating things from a concrete daily actions and tasks being required from a normal system’s everyday use and daily journey perspective. That means we include the very customer, any internal team like engineers working in service desk, an administrative expert plus relationship executives, a product lead and their team the real IT team covering areas such an compliance or network infrastructure departments. These all have diverse usage patterns with sometimes vastly diverse daily requirements. Now is that time to take advantage from these unique opportunities being offered in a timely process improvement initiative?
Then come the more refined aspects: all the expressions to define in concrete terms how “the system must work”, taking into consideration every angle. Functional statements, as requirement, does not only cover the idealistic perfect “happy situation but it must consider less likely scenarios and “exceptions cases”. All of this done in order of improving the very solution.
Every statement when dealing with functional requirement must systematically bring other additional questions too: functional, but mainly based on current user’s activities. A proper inquiry must target on how needs achieving business results, when looked through different perspectives that translates into real day-to-day steps to complete a goal being stated as core objectives. For instance. If part of a overall user journey, you must translate even complex workflows with concrete statement which translate as practical system behaviors along various user operations.
We need also to dig into another very specific requirement’s families: the less visible aspect, typically associated a system’s framework such compliance standards for performance that express real world hard, strict constraints. Or what we often coin as simply rules defining parameters, including regulatory statements, corporate norms, industry compliance benchmarks, or specific or established operating frameworks.
For example, if specific or even highly strict certification is paramount, for instance ISO norm 2022 or PCI-DSS norm of the international card processing businesses. As one example among many. Those frameworks brings their mandatory strict operating guide. And these will form that bedrock for all further development at all levels— whether a new or evolving application , an improvement to existing processes or system or even novel products, and new ventures, ensuring there exists no discrepancy when considering what must be done. And more, many businesses already have an internal guiding principles whether or not these are formally integrated for instance in a internal enterprise architecture manual. It means we must always check that aspect first. This can impact by default and most usually is defined from the outset but in rare events it also will emerge at a late, more mature phase, during one that intake deep dive, during session of brainstorming when the proper inquiry has commenced.
A business or process specialist and also an Architect usually must collect initial set of systemic assumptions and the single best and efficient practical means of doing that properly, starts asking good and pointed specific pertinent questions.
This layered approach to discovering and mapping the technical requirements ensures no stone will be left unturned. Do you wonder how you can reconcile these vastly different perspectives in practical project execution terms?
Let’s address this important question in next episode.
In 2017, Tommy Lee forecasted #Bitcoin would cross the $100,000 mark.
I received an alert: “Bitcoin reaches a new all-time high of $99,261 USD. This is an increase of +47.64% since last month.”
Whether you believe in #cryptocurrencies or not isn’t the point. The point is you must understand what they are and what they represent to make the right decision and fall into the trap of FOMO (fear of missing out).
Contextual Knowledge:
Bitcoin is both a cryptocurrency and the name of its underlying #Blockchain network
To hold Bitcoin, you need a Bitcoin wallet or an online service that manages the wallet (though technically, you aren’t the direct owner)
You can transact in Bitcoin via your wallet or a Bitcoin Visa Card
Bitcoin is considered “digital #gold” – a store of value that historically outperforms other asset classes, with an average yearly return of approximately 182%
Bitcoin’s volatility is unparalleled: -64.3% in 2022, +155.4% in 2023
Bitcoin #ETF (or ETP in Europe) represents the traditional finance-approved version. In 2023, #Blackrock issuing a Bitcoin ETF did send a strong signal
Some businesses are diversifying investments through Bitcoin, with #MicroStrategy holding approximately 331,200 bitcoins as a strategic asset
El Salvador has adopted Bitcoin as its official national currency
Why the current boom?
President Donald Trump, known as a “pro-crypto” leader, has been making strategic moves. After issuing two #NFT collections in 2022 and 2023, he launched “The Defiant Ones” #DeFi platform with his sons Eric and Donald Jr. Trump.
Musk, notorious for supporting the #Dogecoin community, has seen the cryptocurrency soar by 193.44% in a month, with a pivotal moment on November 5th, 2024 – the US election day.
Markets aren’t always rational, but the sequence of events always tells a story.
Golden Rule: If you do not understand the product, do not invest.