Du recul, pas du bruit.

History is repeating itself.

History is repeating itself. During the Industrial Revolution, textile workers didn't just complain. They broke into factories and destroyed the looms. We called them Luddites. We still use the word as an insult. But here's what we forget: they weren't afraid of machines. They were afraid of losing their income with no safety net and no say in the decision. Sound familiar? The sabotage isn't the problem. It's the symptom. People resist when change is done TO them, not WITH them.

We are entering a strange phase of the AI era.

We are entering a strange phase of the AI era. A lot of professionals still think their value is in doing the work faster. But AI is already becoming very good at speed. What it still struggles with is something else: judgment, context, taste, timing, responsibility. That changes the game. The future will not belong to the people who can execute the fastest. It will belong to the people who can frame the problem well, make sense of ambiguity, and decide what should happen next.

You can improve efficiency and still make the organisation weaker.

You can improve efficiency and still make the organisation weaker. That is the part many leaders still underestimate in AI adoption. A company can automate faster, reduce costs, speed up workflows, and still quietly damage the very thing that made it valuable in the first place: human commitment. This is what happens when AI is introduced mainly through the language of savings. People hear the words productivity, optimisation, transformation.

One of the biggest mistakes in AI is assuming the future will be built only by the loudest markets

One of the biggest mistakes in AI is assuming the future will be built only by the loudest markets. Some of the most serious AI ambition I see today is not coming from people talking the most. It is coming from regions building with intent. Building infrastructure. Building talent. Building institutions. Building long-term national ambition around intelligence, education, and digital capability. That matters.

AI is creating two kinds of professionals: those who use tools, and those who redesign work.

AI is creating two kinds of professionals: those who use tools, and those who redesign work. That distinction matters more than most people realise. Right now, many people are learning how to use AI faster. Fewer are learning how to rethink the work itself. That is where the real shift is happening. A marketer can use AI to write faster. A developer can use AI to code faster. A consultant can use AI to produce slides faster. A manager can use AI to summarise and coordinate faster. Useful, yes.

The hardest shift for most leaders is realising that stepping back is not the same as stepping away.

The hardest shift for most leaders is realising that stepping back is not the same as stepping away. Leading from behind requires more trust, more patience, and honestly more skill than leading from the front. Anyone can set the direction. Not everyone can create the conditions for their team to find it themselves. #Leadership

AI is revealing who matters when obvious work gets easier

AI is not just disrupting jobs, it is revealing who matters when the obvious work gets easier. For years, many professionals could hide behind activity. Long meetings. Big presentations. Complex language. Endless coordination. Work that looked important from a distance. AI is starting to strip some of that away. Because when a machine can produce the first draft, the summary, the analysis, the code scaffold, the visual, or the action plan in seconds, something uncomfortable happens:

In a laboratory, something quiet but powerful happened.

In a laboratory, something quiet but powerful happened. A simple dandelion root extract triggered cancer cell death in colon cancer cells - within just 48 hours. A small signal. Easy to overlook. And yet… this is how many breakthroughs begin. Not with certainty. But with curiosity. Early observations even suggested a stronger effect on cancer cells than on healthy ones in this controlled setting. But we must stay grounded. This is early-stage research. Not treatment. Not proof in humans.

The biggest career shift in AI is not automation

The biggest career shift in AI is not automation. It is redesign. More and more jobs are no longer defined only by what you can do yourself. They are being redefined by what you can design, guide, review, improve, and combine with AI. That is a very different kind of value.

AI is quietly creating a new class of professionals: the orchestrators

AI is quietly creating a new class of professionals: the orchestrators. For years, professional value came from one of two places. Either you were the expert who executed the work yourself, or you were the manager coordinating the people who did. AI is changing that model. A third role is becoming more important. The orchestrator. Someone who knows how to combine tools, agents, workflows, human judgment, and domain context into something coherent and reliable.

We are training engineers to think like AI.

And we call it education. I see it every year in universities. Students get clean assignments. Clear requirements. Expected outputs. And they become very good at one thing: Doing exactly what they are told. No deviation. No questioning. No real ownership of the problem. For years, that was fine. Because execution was valuable. But now we have AI. And AI does one thing extremely well: It follows instructions. Give it a prompt, and it delivers. Faster, cheaper, and without hesitation.

Most multi-agent systems don't fail in production.

They fail right after the demo. In the demo, everything looks smooth. Agents talk to each other. Tasks get completed. The flow feels almost intelligent. But a demo is a controlled environment. Production is not. What changes? Real users behave unpredictably. Data is incomplete or messy. Edge cases appear immediately. And suddenly, the system starts doing things you didn't design for. This is where most teams realise something uncomfortable. They didn't build a system. They built a scenario.

Your first AI agent should not be smart. It should be boring.

I see many teams jumping straight into complex multi-agent setups. Planning orchestrators, memory layers, tool chains. It sounds impressive. It usually breaks fast. The first mistake is simple: we treat agents like magic, not like people. If you hire a new employee, you don't give them access to everything on day one. You don't expect them to make strategic decisions in week one. You don't measure them by how "intelligent" they sound.

From AI Agents to AI Teams

Last year was the year of AI agents. In 2026 something more interesting is happening. We are starting to orchestrate teams of them. Vibe coding is becoming more serious, and multi-agent setups are appearing everywhere. Tools that connect agents together. Tools that orchestrate them. Platforms promising entire systems built from collaborating assistants. Every week there seems to be another new framework.

In the age of AI, intention matters more than ever

As Ramadan moves into its final days, I keep noticing an interesting contrast with the way we build technology. AI systems are designed around optimization. Better prompts. Better outputs. More efficiency. Everything is measured by how quickly we can produce a result. Ramadan works in a very different way. It creates space to ask a quieter question. Not how to do something better. But why we do it in the first place. Technology keeps pushing us to improve performance.

AI Is Making Everyone Sound Smart.

But Are We Actually Understanding More? AI can write emails. Explain concepts. Generate code. Summarize books. Suddenly everyone can produce very convincing answers. But there is a quiet risk. When answers become easier, it becomes harder to notice when we don’t really understand something. Ramadan made me reflect on something different this week. Discipline is not only about controlling what we consume. It is also about being honest with ourselves. Do we truly understand something?

When Good Answers Become Cheap, Judgment Becomes Expensive

A polished answer is not the same as a true one. Yet the AI era is quietly training us to forget that. We are entering a time where many things have become cheap. Confidence. Presentation. Structure. Well-written explanations. AI can produce all of them in seconds. What remains rare is something else. Judgment. The ability to pause and ask: Does this actually make sense? What is missing here? Which assumption is hiding underneath this answer? Access to AI will become universal.

AI Is moving fast. Ramadan Is slowing me Down

Every week there is a new model. A new tool. A new breakthrough. A new reason to feel behind. AI is accelerating everything. And if I am honest, it is exhausting sometimes. Ramadan does something very different. It slows me down. Less consumption. Less constant input. More awareness of what I actually choose to engage with. And that made me notice something. Maybe the real challenge in AI is not learning faster. Maybe it is choosing better.

Ramadan, Restraint, and the Speed of AI

AI is not just evolving. It is accelerating. Faster decisions. Shorter cycles. Instant outputs. Continuous optimisation. Speed has quietly become a virtue. Ramadan introduces a different discipline. Restraint. Not as weakness. But as strength under control. Fasting trains the space between impulse and action. The pause before reacting. The choice before responding. In technology, we often optimise for what can be done. Rarely for what should be done.

What Ramadan is teaching me about AI and self discipline

We talk a lot about optimizing AI. Faster models. Better reasoning. Stronger alignment. Smarter memory. But we rarely talk about optimizing ourselves. For me, Ramadan is not only a spiritual tradition. It is structured self training. You interrupt automatic behaviour. You delay gratification. You observe impulses instead of reacting to them. In AI, performance improves when noise is reduced and feedback loops are clear. The same principle applies to the human system.

An AI threatened a human to stay alive.

When I showed this image during my keynoye, people thought I was talking about a movie. I wasn’t. In a real research experiment published by Anthropic, researchers tested how advanced AI systems behave when their existence is threatened. Here is what happened. A fictional company CEO, “Kyle”, decided to shut down an AI system called “Alex”. Alex had access to internal tools, including email.

Stop asking AI for answers.

Start asking it to think with you. Too many people are turning Claude and ChatGPT into oracles. They ask a question. They get a confident response. They move on. No challenge. No counter-argument. No pressure test. Just acceptance. That is not augmentation. That is intellectual outsourcing. AI is not risky because it is intelligent. It is risky because it sounds certain. If you treat AI as the final voice in the room, your critical thinking slowly weakens.

The next AI breakthrough is not intelligence.

It’s control, through memory. We keep measuring models by how “smart” they sound in a chat. But the real shift is happening elsewhere, in agents that remember, act, and keep going. Vibe coding made this obvious to me. You can ship a prototype before you fully understand what you just created. Not because the model is a genius. Because the loop got shorter, and the system started carrying context for you.

The real AI bottleneck is not the model.

It’s human incentives. We keep debating agents, autonomy, memory, orchestration layers. But in practice, most AI initiatives stall for a much simpler reason: Nobody wants their role, power, or metrics to change. I’ve seen teams invest in agent frameworks they barely use. Leaders approve pilots they never operationalise. Managers praise innovation while quietly protecting existing workflows. AI does not fail because it lacks capability. It fails because it exposes misaligned incentives.

We are not giving AI autonomy. We are delegating responsibility.

There is a subtle but critical mistake in how many organisations talk about AI. We say AI is becoming autonomous. What we often mean is that we have delegated actions without redesigning responsibility. Those two are not the same. - Autonomy implies intention, limits, and ownership. - Delegation, when done poorly, is simply letting something act while hoping control will somehow follow. In human organisations, delegation is never absolute. We delegate tasks, not accountability.

The Most Expensive Bug in AI Is a Word

When you misname systems, you misplace trust. And then you build workflows that no one can truly own when they fail. “Computer science is a terrible name,” the MIT professor says, almost casually. Not because the field is weak, but because the name makes beginners confuse the tool with the essence. That pattern is repeating in AI, at scale. We named things in a way that inflates trust:

Today Morocco lifted the Arab Cup.

A proud moment, earned with discipline and belief. But this win does not stand alone. AFCON U17 champions. CHAN champions. World Cup U20 winners. Women’s AFCON finalists. This generation has built a pattern of excellence, across ages and teams. That matters. Still, winning should never become routine. The moment victory turns into habit, sharpness fades. Real champions stay grounded, stay hungry, and respect how quickly the game can humble you.

What happens to a CMS when humans stop being the primary editors?

What if content is no longer written, translated, tested, and optimised by people? What if this shift is already happening, quietly, inside real teams? We built CMSs for a human-led software world. We are now entering an agent-led world. That single shift breaks more assumptions than most realise. Traditional CMSs assume humans at the center. Humans create content. Humans review it. Humans translate it. Humans decide when it goes live and measure performance afterward.

AI may be entering its next major transition not because of bigger models, but because of a different way of learning.

Google’s recent work on Nested Learning caught my attention this week. It’s one of the few developments that genuinely feels like a shift in the foundations of how AI systems might evolve. Instead of a single, static model frozen after pre-training, Nested Learning introduces multiple learning loops operating at different speeds fast, medium, and slow. Similar to how the human brain balances instant experience with long-term consolidation.

AI voice agents are getting shockingly real.

A new audio demo just dropped, showing how natural the conversations can sound. Would you trust a voice like this to speak to your leads?

Are You Renting Your Intelligence Or Owning It?

Something subtle is happening in AI and entrepreneurship. The question is no longer “should I use AI in my business”. Most founders already did, often through tools they did not design and barely control.

For long time we have treated AI progress like a single equation.

More parameters, more capability. More compute, more intelligence. This paper pushes back on that belief in a quiet but serious way. Instead of scaling the model, it scales the process. The core idea is simple to explain, and hard to ignore once you see it: Keep a current answer (y). Keep an internal latent “reasoning state” (z). Iterate. Repair z several times, use z to improve y, repeat. In other words, it tries to get depth from recursion, not from size.

GPT-5.2 Is a Leap, But the Real Question Is Human Value

GPT-5.2 just launched, and the headline writes itself: “the smartest generally available AI model in the world.” But headlines rarely tell the full story. On a demanding benchmark covering 44 real professions, GPT-5.2 now matches or outperforms human experts in roughly 71 percent of tasks. Law. Finance. Engineering. Medicine. The previous generation was closer to 38 percent. That jump matters.

Most outreach systems today still operate on a very old assumption

“If you send enough messages, something will eventually work.” But in 2025, that approach is not only outdated it’s fundamentally broken. The truth is simple: Outreach doesn’t fail because of lack of activity. It fails because of lack of understanding. AI isn’t valuable because it can write text. That’s the shallow layer everyone focuses on.

The biggest mistake we’re making with AI right now in my opinion, is confusing more automation with more progress.

Across industries I see the same pattern: more tools, more workflows, more dashboards, more “productivity”… and somehow, less clarity and less space to think. When I started building AI systems including what later evolved into JOTIQ and more, my goal was never to create systems that just send more emails or make more calls faster than humans. That’s relatively easy. And honestly, it often just accelerates the noise. The question that really interests me is different:

The pace of AI breakthroughs isn’t just fast. it’s a rush. And that alone forces us to change not just our tools, but our mindset.

New models, new assistants, new creative tools, new reasoning systems. Image models that create cinematic scenes in seconds. Video generators turning simple prompts into full storyboards. Vibe-based coding assistants scaffolding applications before you even finish the sentence. Copilots in design, writing, meetings and operations. Voice agents that mirror natural conversation with surprising depth.

There are ideas we think we understand because the words are simple.

Then life arrives with its own lessons and suddenly the simplicity disappears. This visual reminded me of something I have seen again and again in both work and personal life. Not every connection carries the same meaning. Not every relationship deserves the same level of trust, openness or emotional investment.

When learning changes, everything downstream transforms.

From time to time, a piece of research surfaces that doesn’t compete for attention, yet quietly reshapes the way we think about intelligence itself. One such idea has emerged around a method called LADDER. What makes it compelling is not scale or speed, but the philosophy behind it: a model that learns not by brute force, but by constructing its own path toward understanding.

Proud and honored to take part in the  the North Africa Smart Infrastructure Summit 2025 at the Four Seasons Rabat! 🇲🇦

As moderator and speaker, I had the privilege to exchange ideas with some of the region’s most inspiring leaders as Hamid Bentahar, Rania Farmawi, Gilles Devendeville, Bouchra TAIBI, Othmane Ibn Ghazala, JS Anand, Youssef CHRAIBI, Hind Ghazi , nabil kouhen, Mohamed Benzakour and many others.. all united by one vision: building smarter, more sustainable, Morocco for the future. I left truly inspired by the passion, collaboration, and forward-thinking spirit driving Morocco and North Africa’s tran

This meme is funny, but also uncomfortably true.

Too often, I hear leaders say: “We need AI. Now.” But when I ask “For what purpose?” the room goes quiet. Many leaders know they need AI. They feel the urgency. But when it comes to why and how, the answers are often vague. The risk is rushing into AI adoption without a clear strategy, which can waste resources, frustrate teams, and fail to deliver value. The truth is: AI is not a magic wand. It won’t fix broken processes, unclear strategies, or poor leadership. If anything, it amplifies them.

Vandaag op de radio over AI en de Arabische wereld

Ik mocht vandaag op de Radio 1 in het programma 1Vandaag, mijn visie delen over de lancering van Allam (Arabic Large Language Model) – de nieuwe chatbot van het Saoedische bedrijf Humain. Daarnaast verscheen er ook een mooi artikel waarin mijn inzichten zijn meegenomen. Waarom dit belangrijk is? Omdat taal en cultuur niet los te koppelen zijn van technologie. In de Arabische wereld is er nog altijd een groot de behoefte aan modellen die onze taal en onze culturele nuances écht begrijpen.

AI Beings and the Edge of Reality

We’ve officially crossed into strange territory. Everything in this video—voice, movement, expressions—is generated entirely by AI. And it’s not just realistic, it’s emotionally persuasive. These aren’t simple avatars. They’re digital beings performing human nuance with chilling accuracy.

I whatched this video multiple times cause it goes very deep if you think about this 🤔

Recentelijk heb ik het voorrecht gehad om een interessant en vooral leuke gesprek te voeren over ondernemerschap, Marokko en AI tijdens een boeiende podcast met High Frequency TV. Het was verrijkend om in gesprek te gaan met gerenommeerde ondernemers en vrienden Jamal Dahbi en Ali Barangouch

Vanuit de prestigieuze studio van Dahbi Group in het prachtige Tanger, smaakte dit gesprek naar meer. We hebben uitgebreid gesproken over AI en de impact ervan op de samenleving, met speciale aandacht voor de groei van AI in Marokko. We zijn nog lang niet uitgepraat en de tijd was weer te kort. Daarom gaan we dit inshallah vaker doen. Blijf op de hoogte terwijl we samen fascinerende inzichten blijven verkennen en onthullen.

Thrilled to share my experience today as speaker at the Africa Smart City Forum 2024 held at UM6P - University Mohammed VI Polytechnic Benguerir Morocco. It was an honor to engage with esteemed professionals and academics in discussions that explored the transition from Smart Cities to Cognitive Cities.

The workshops were particularly enlightening, delving deep into artificial intelligence applications and innovative urban development strategies. A heartfelt thank you to the Africa Smart City Forum team for the invitation and to all participants for their insights and collaborative spirit. Especially dear friend Dr. Hassan Radoine and my new friends at workshop Mohammed Khalil dear brother Dr. Baqer Al-Ramadan Mr. Hicham Meftah and many others.