Introduction
Entrepreneurial doctrines are seductive as long as they explain other people’s victories; they become genuinely interesting when they must answer for the fractures of the person who formulated them. For years, Oussama Ammar helped shift the boundaries of French entrepreneurial discourse: making closely guarded codes accessible, legitimising ambition, and teaching a generation that failure was not necessarily a fault. Then The Family broke apart. Conflicts between partners, legal proceedings and a 2025 British ruling finding him civilly liable for fraud and breaches of fiduciary duties—a ruling whose circumstances and interpretation he disputes—changed the framework within which his words can now be heard.
This interview seeks neither to erase that rupture nor to confine his thinking within it. Ammar sets out what he has rebuilt intellectually after The Family: a more selective, asymmetric and personal doctrine of company-building, in which access can be democratised but talent cannot; in which value is acknowledged as asymmetric; and in which conviction requires an identifiable person to assume responsibility, while money and equity must be governed by an architecture of control.
The entire tension lies in that balance. How can decision-making be concentrated without turning one’s own judgement into the system’s central risk? How can entrepreneurial risk be defended without diluting responsibility? How can one think against the consensus without mechanically preferring the opposite narrative? And what remains of the mentor when AI can memorise, compare and detect certain patterns better, if not the part no machine yet assumes: putting one’s reputation, capital and several years of one’s life behind a decision?
Ammar articulates strong convictions here about talent, Europe, failure and power. Yet his most decisive answer may be the one in which he agrees to define his own breaking point. A body of thought does not prove its strength through the confidence with which it is expressed, but through the precision with which it identifies what could disprove it. What remains is to strip away the persona, the legend and the controversies, and allow his new doctrine to face the only opponent it fully recognises: reality.
DEMOCRATISE OR SELECT?
JLP Décryptage
For a long time, The Family championed the idea that an entrepreneur could emerge from anywhere. Today, your model appears far more concentrated: you say that you build around five ventures a year and explicitly seek people you describe as “singular”. Your selection protocol even relies on one hundred questions assessed almost entirely by artificial intelligence.
With hindsight, what can truly be democratised in entrepreneurship: access to tools and capital, or the ability to build an exceptional company?
And have you changed your mind about the scarcity of entrepreneurial talent?
Oussama Ammar
What I have changed my mind about most is what the word “democratise” meant. At the time of The Family, Europe’s problem was obvious: information, codes, networks and capital were controlled by a small caste. Many capable people could not even get into the room. So we opened the door. It was necessary, and I have absolutely no regrets about doing it.
But opening the door does not make everyone exceptional. Access to tools, knowledge, models, capital and now almost infinite intelligence through AI can be democratised. Desire, endurance, speed of learning, the ability to withstand ten years of uncertainty, or that rather strange form of obsession that drives someone to continue when every reasonable argument tells them to stop cannot be democratised.
I do not believe entrepreneurial talent has become scarcer. I believe we measured it badly. We often confused talent with qualifications, eloquence, networks and conformity with the prevailing codes. This led me to a very personal experiment in using AI. When I look for people to build with today, I no longer begin with a thirty-minute interview in which my impression can be dominated by charm, pedigree, accent or the ability to tell a compelling story. I ask them to answer one hundred questions.
These questions are not looking for the correct academic answer. They explore several dimensions: general knowledge, pattern recognition, reasoning when information is missing, the speed with which someone turns an idea into action, and their relationship with risk, failure, conflict and learning. Some answers seem ordinary in isolation. Their value comes from the whole. A person may describe themselves as highly ambitious in one answer, then reveal twenty questions later that they systematically organise their life to avoid any situation in which they might lose. That discrepancy is more instructive than the initial declaration.
I then use AI as a reading and comparison machine. It can hold all one hundred answers in memory at once, look for consistencies, contradictions, repetitions and weak signals, and then produce a structured analysis. Where a human being mainly remembers the three most brilliant answers, it can connect answer 7 with answer 63 and answer 91. It does not tire, fall in love with a turn of phrase, or mechanically assign greater value to the candidate who resembles founders it already knows.
When I say AI performs almost the entire assessment, I do not mean that an algorithm automatically chooses my partners. I mean that it does the heavy work of reading, connecting, challenging and conducting the initial analysis. My involvement is focused on profiles that resist a simple conclusion: those who display rare signals but also significant contradictions, or those whose potential appears real without fitting neatly into a category. AI does not assume final responsibility. It helps me avoid confusing my intuition with the truth.
The truth is less comfortable than the slogan: potential is widely distributed, but the ability to build an exceptional company is not. Democratising entrepreneurship does not mean promising that everyone will win. It means ensuring that birth, address or personal connections no longer determine in advance who has the right to try.
POWER AS ARCHITECTURE
JLP Décryptage
The Family was built around a collective organisation: partners, teams, investors and investment vehicles. Your current model is almost the reverse: a handful of ventures each year, no traditional fund, and your direct involvement as a co-founder.
Is this simply a change of business model, or has your experience led you to a more radical conviction: the earlier a company distributes power and responsibility, the more it creates the conditions for its own fragility?
And if you now concentrate decision-making more heavily, what protects you from the symmetrical error of making your own judgement the system’s primary risk?
Oussama Ammar
I will answer more radically: no, I am not looking for a system in which someone can tell me no on the very matter that constitutes my specific expertise. I am looking for a system in which responsibilities correspond to the reality of value creation. The Family was unbalanced because we presented as equal an organisation that was not equal in practice.
By the end, I represented 89% of The Family’s NAV. That figure does not mean I was right about everything, or morally or intellectually superior to my partners. It says something far more specific: in the work of finding, winning and supporting the deals that created the portfolio’s value, my contribution was overwhelmingly dominant. A healthy organisation should have begun from that reality. We preferred to preserve the fiction that three partners were essentially producing the same thing.
There is a fairly simple test today. Since I left, Alice and Nicolas have made no significant new deals. I still do, despite the controversy, despite the destruction of my reputation, and despite the fact that anyone who works with me accepts an immediate social cost. If the value came primarily from The Family brand, its team or our collective complementarity, the outcome should have been the reverse. It is not.
That does not mean I am the best in the world. There are obviously investors, founders and dealmakers far better than me. But the best are precisely very difficult to pair with people as good as they are, because the nature of the role is contrarian. To produce an exceptional deal, you have to see something before others do, believe in it more strongly than they do, and be willing to stake your reputation when the consensus says no. Two people can collaborate on the analysis, but at the decisive moment there generally has to be one conviction, one signature and one accountable person.
This is where the comparison with Beyoncé and Destiny’s Child is useful. Destiny’s Child was a real group. The other members were talented and contributed to its success. But with hindsight, no one seriously claims that artistic and commercial value was distributed equally. Beyoncé was not one third of the group. The group was the temporary architecture within which Beyoncé became Beyoncé. Recognising that does not humiliate the other members; it simply describes the actual distribution of talent and value.
Venture capital and venture building are not industrially scalable professions. There is no need to do hundreds of deals. In reality, a few decisions account for almost all of a portfolio’s performance. Not all deals are equal: one can repay ten years of mistakes, while fifty mediocre deals merely create the illusion of activity. The essential skill is therefore not operating a volume machine. It is recognising the rare moments when time, money and conviction must be concentrated.
My current model embraces that asymmetry. I build few ventures, become directly involved as a co-founder and take responsibility for the choice. The mechanism that protects me from my own error is not a committee tasked with neutralising my conviction. It is reality: my money, my time, my reputation, customers, revenue, value created, and the ability of companies to survive without me. I want absolute controls over money, equity and fiduciary obligations; no one should be able to act alone in those areas. But when it comes to entrepreneurial selection and initial conviction, organising an internal veto would amount to removing precisely the skill the system is designed to exploit.
AFTER THE BREAKDOWN OF TRUST
JLP Décryptage
In 2025, a British court found you civilly liable for fraud and breaches of fiduciary duties towards The Family, a ruling whose circumstances and interpretation you have publicly disputed.
Irrespective of that disagreement, an entrepreneurial question remains: after a breakdown of trust on that scale, what have you changed in concrete terms about the way you govern money, equity and relationships between partners?
In other words, if a new co-founder told you today, “I do not want to have to trust you; I want a system that makes that trust unnecessary,” what mechanisms of control, transparency or countervailing power could you show them?
Oussama Ammar
I dispute that ruling, the circumstances in which it was obtained and a significant part of the narrative drawn from it. But disputing a judgment does not entitle me to ignore the question it reveals. When a breakdown of trust reaches that scale, saying “trust me” is no longer a serious answer. Fundamentally, it should never have been one.
What I have changed comes down to one idea: turning implicit relationships into explicit systems. A company’s money must remain in its accounts. Every flow of funds must have a justification, supporting documentation and an accountable person. Sensitive transactions must require multiple approvals. Cap tables, agreements, economic rights and commitments must be maintained by independent professionals and accessible to the people concerned. Partners must receive regular reporting, be able to consult source documents, and have alert channels that do not pass through the person they want to scrutinise.
I also separate far more clearly four things that startups like to mix together: friendship, capital, power and work. People can be friends without having the same economic rights. Someone can be a shareholder without running the company. Someone can run the company without being able to move money alone. And an oral promise, however enthusiastically made, is not governance.
To the new co-founder who told me they wanted a system that made trust unnecessary, I would reply: you are right. Trust should allow us to move faster in areas that a contract cannot anticipate; it must never replace basic controls. I cannot promise them the absence of conflict. I can show them an architecture in which neither of us needs to be a saint for the company to remain protected.
THE PRICE OF CONTRARIANISM
JLP Décryptage
You once enjoyed substantial institutional legitimacy: in 2018, for example, you took part in the European Union’s Global Tech Panel. A few years later, in December 2023, you relayed on Sans Permission a story claiming that artificial intelligence had established that images from the US Moon missions were fake and that NASA had acknowledged it—a story whose factual elements were subsequently debunked.
An entrepreneur must often know how to think against the consensus. But how do you now distinguish productive contrarianism, which makes it possible to see before others do, from the simple pleasure of believing a story because it contradicts the dominant narrative?
In other words, what is your own protocol for determining whether you are thinking differently—and not simply getting it wrong?
Oussama Ammar
The premise of the question is already wrong. It comes from an AFP article whose journalist clearly did not listen to the entire episode. He isolated a story from the analysis that accompanied it, then presented that story as my conclusion. That is precisely the kind of mechanism I was criticising: an authority produces a short version, other media reproduce it and, in the end, the fact that AFP wrote it becomes proof that AFP was right.
But let us go further: in that episode, whether the initial story was true or false was not the most interesting issue. A story can be factually false and reveal something profoundly true about why millions of people want to believe it. Intellectual work does not consist solely in stamping “true” or “false” on a claim. It consists in understanding what the existence and circulation of a narrative teach us about the society that produces it.
Take flat-earthers. A flat-earther is generally defined as someone who believes the Earth is flat. That is accurate, but superficial. Above all, a flat-earther is someone who has lost trust. They no longer believe scientists, journalists, schools, images, institutions or governments. And the system responds to that loss of trust by repeating more loudly: “You must believe us, because we are the experts.” It is a circular answer. For someone who no longer recognises official authority, it turns every official intervention into further evidence of the conspiracy.
The real problem begins at school. Most of us do not learn that the Earth is round by demonstrating it. We learn it because a teacher, a book or an institution asks us to trust it. We replace one belief with another that is better certified socially. Yet we could learn to observe the Earth’s shadow, the gradual disappearance of a boat below the horizon, the difference between shadows in two locations, or reproduce Eratosthenes’ reasoning. At that point, we no longer need to trust the teacher. We possess a method for verifying what they claim.
That is what the system should teach: not trust in the system, but the ability to question any assertion, including the system’s own. A robust institution should not say, “Believe me.” It should say, “Here is how to prove me wrong.” If we truly taught demonstration, doubt would cease to be the enemy of knowledge. It would become its driving force.
An entrepreneur must operate in this way. They cannot wait for an authority to certify that a market exists or that a technology will work. If they have to trust the consensus, they are already too late. But thinking against the consensus does not mean arbitrarily choosing the opposite narrative: it means turning an intuition into an experiment, looking for what could refute it, building a product, speaking to customers and allowing reality to answer.
Nor should they fear error. That fear produces people who delegate their judgement to those who hold a title. Error that is acknowledged, tested and corrected is the normal price of learning. The danger is not having believed a false hypothesis; it is having no mechanism for discovering that it was false. Good contrarianism therefore does not consist in placing more trust in alternative narratives than in official ones. It consists in constructing a way of thinking that needs to trust as few people as possible.
FAILURE OR RESPONSIBILITY?
JLP Décryptage
For years, you helped normalise entrepreneurial failure in France: a startup can disappear, a product can fail and a conviction can prove to be wrong without condemning the person who undertook the venture.
But a ruling concerning fiduciary duties does not belong to the same category as a startup failure.
Where do you now draw the line between the legitimate risk an entrepreneur must accept and the personal responsibility that cannot be absorbed into the narrative of entrepreneurial failure?
And is that line exactly the same for you today as it was when you taught entrepreneurship at The Family?
Oussama Ammar
We need to stop kidding ourselves about the supposed right to fail in France. We spend our time explaining that entrepreneurial failure is now accepted, then organise every bankruptcy as an investigation designed to find someone to blame. Why does France produce so much litigation whenever a company fails? Why were 100% of The Family entrepreneurs who failed with Fraca threatened? Because we do not know how to do what every innovation economy must know how to do: recognise a loss, write it off and move on.
In France, a mistake is never truly settled. It must be explained for ten years, retried using information acquired after the event, and then transformed into a moral failing. As long as the company is growing, a risky decision is called vision. The day it fails, exactly the same decision becomes recklessness, mismanagement or misconduct. That is morality written by the outcome, not a serious system of responsibility.
The question of fiduciary duties imports into the startup world a logic drawn from wealth management and investment funds. But a startup is not a fund entrusted with preserving capital. It is a machine created to consume capital in order to discover something uncertain: a product, a market, a technology or a distribution model. The money is not locked in a safe awaiting a predictable return. It is converted into salaries, code, prototypes, failed hires, campaigns that do not work, and hypotheses that reality destroys.
Everyone knows the statistical distribution when they invest. Most startups will fail. A few will repay the others. Investors accept that asymmetry because they want access to the upside. They cannot invoke venture logic when a company wins, then turn into prudent, outraged creditors when it loses. You cannot privatise the upside and litigate every downside.
Of course there is a line. Lying in the accounts, deliberately concealing decisive information or taking the money for yourself is not entrepreneurship. But when money has been committed to building the company, when investors know that they are financing a hypothesis, and when failure comes from the market, the product, the timing or a decision sincerely taken to save or develop the company, it must be called what it is: a loss. Not an ex-post fiduciary fiction invented to guarantee professional investors that they will never lose.
The problem with applying the concept of fiduciary responsibility to startups without nuance is that it always makes it possible to manufacture fault retrospectively. A startup survives through precisely the kind of decisions that no reasonable manager of a mature asset would make: concentrating resources on an unproven technology, hiring ahead of revenue, pivoting abruptly, investing cash in growth that remains hypothetical. If a judge analyses those choices as the choices of an administrator charged with preserving an asset, the entrepreneur is guilty by definition as soon as the bet fails.
The real boundary therefore does not lie between entrepreneurial risk and fiduciary responsibility, as though the latter automatically possessed moral superiority. It lies between deception and an acknowledged bet. Deception must be punished. An acknowledged bet must be allowed to lose. If we reject that distinction, we are not protecting the economy; we are merely selecting entrepreneurs who have nothing left to lose, those with enough money to defend themselves for ten years, or those who leave to build elsewhere.
Is that exactly what I taught at The Family? I already taught that a startup could die without its founder being a criminal or an idiot. But I did not go far enough. I thought changing the vocabulary would change the culture. France has learned to celebrate failure on stage while continuing to punish it in the courts, banks, press and relationships between investors and founders. A right to fail that does not include the right to a write-off is not a right. It is a conference.
IF AI SELECTS TALENT, WHAT PURPOSE DOES THE MENTOR STILL SERVE?
JLP Décryptage
Your “Singularity” protocol asks candidates to answer one hundred questions concerning, among other things, their abilities, culture, pattern recognition and capacity for action. You state that artificial intelligence performs 99.99% of the assessment, with your personal involvement reserved for borderline cases.
This is an interesting reversal: for a long time, part of an investor’s or mentor’s value lay precisely in the human ability to recognise an exceptional founder before anyone else.
If a machine becomes better than we are at identifying the people capable of building tomorrow’s companies, what part of your own value to a founder do you regard as genuinely irreducible to AI?
Taste? Judgement? Network? The courage to make a decision? The responsibility of being wrong?
Oussama Ammar
First, I need to explain what I am actually doing, because it may sound more mysterious than it is. I have a very concrete problem: I want to meet exceptional people without limiting my search to those who already know how to access my network or perform well in an interview with me. So I use AI to assist me in the first part of the selection process.
The candidate answers one hundred questions. They concern their abilities, culture, pattern recognition and capacity for action, but also how they think when there is no obvious answer. I am not looking for one hundred opinions aligned with mine. I am looking for a mental structure. Does the person connect distant fields? Do they distinguish a fact from an intuition? Do they change their mind when new information appears? Do they produce concrete examples or merely seductive abstractions? Do their answers describe someone who actually takes action, or someone who knows the vocabulary of action very well?
The value of AI is not that it magically possesses a definition of genius. Its advantage is far more ordinary and far more powerful: it can read everything. A human conducting an interview retains a voice, an energy, two anecdotes and an overall impression. AI can compare all one hundred answers with one another. It can detect that a person uses the same reasoning structure throughout, under different formulations. It can also note that a candidate claims intellectual independence but systematically seeks the group’s approval whenever a decision carries a personal cost.
I therefore ask it to do several things: summarise the candidate’s mode of reasoning; identify their strongest capabilities; isolate contradictions; distinguish assertions from evidence provided in the answers; detect signals of speed, endurance, curiosity and action; then formulate the strongest case for the candidate and the strongest case against them. This last part matters. AI used solely to confirm an intuition does not remove human bias. It industrialises it.
That is what an assessment performed almost entirely by AI means in my case: it carries out the exhaustive reading and initial analysis that I would be unable to reproduce with the same attention across hundreds of people. I intervene personally only when the profile is ambiguous, when rare qualities coexist with significant risks, or when the analysis produces a conclusion that deserves to be challenged. I do not delegate the moral decision of choosing a partner. I delegate the part of the work in which memory, comparison and discipline matter more than my ego.
Above all, this experiment has taught me that the mentor no longer has much value as an oracle. If a machine can memorise, compare and detect certain patterns better, I am not going to invent a mystique of human instinct to protect my status. My value begins after selection: turning potential into decisions, choosing a problem, building a strategy, recruiting the first people, winning the first customers and deciding when to persist or give up.
Taste, judgement and even networks will increasingly be assisted by AI. What remains irreducible, for now, is commitment. A machine can recommend a decision; it does not put its reputation, money, relationships and five years of its life behind that decision. It does not bear the social and moral responsibility for error. The mentor who merely supplies answers will disappear. The one who remains useful is a co-founder capable of creating the conditions for action and remaining present when a recommendation becomes a consequence.
WHAT EUROPE FAILED TO UNDERSTAND
JLP Décryptage
You experienced the European ecosystem from within: The Family, French entrepreneurs, investors and then the European Union’s Global Tech Panel. You now build your projects from Dubai and take a far more critical view of European institutions, regulation and entrepreneurial culture.
You have therefore occupied both positions in succession: that of someone seeking to transform the European ecosystem from within, and that of someone who chose to build elsewhere.
With that dual experience, what did you misunderstand about Europe when you were still trying to change it?
Conversely, what do those who leave it for Dubai or the United States misunderstand when they believe Europe’s problem can be reduced to taxation and bureaucracy?
Oussama Ammar
What I misunderstood was that Europe was not suffering solely from a lag in public policy. I thought that with more capital, better rules, a few visible successes and a new generation of entrepreneurs, the ecosystem would mechanically converge with Silicon Valley. I underestimated the cultural depth of the problem.
Europe likes innovation as an outcome, much less as a process. It wants technology companies once they are safe, clean, profitable, well governed and compatible with existing balances. But disruption almost always begins as something uncertain, inelegant and disproportionate. You cannot want the consequences of boldness while organising every institution to avoid the conditions that make it possible.
I had also overestimated our ability to change a system through discourse alone. An ecosystem is a reward machine. It reveals what it wants through the careers it secures, the fortunes it permits, the failures it forgives and the speed with which it decides. As long as the incentives remain the same, conferences about innovation are primarily a form of theatre.
But those who leave are also mistaken when they reduce Europe to taxes and forms. Europe possesses exceptional historical, cultural, scientific and human depth. It teaches nuance, provides access to remarkable talent and offers a quality of life that many rediscover after leaving. Dubai is not magical; it is a society that embraces growth, speed, ambition and its openness to foreigners who contribute to its project more openly. The United States has its own social violence and its own conformism.
The real issue is therefore not to determine which territory is morally superior. It is to know which environment rewards the project you want to build. I have stopped trying to convince Europe to become something else. I build where the gap between stated ambition and actual behaviour is currently the smallest.
WHAT COULD PROVE YOU WRONG?
JLP Décryptage
Your current model rests on a strong thesis: very few ventures, “singular” individuals, a structure without a traditional fund, direct involvement as a co-founder and intensive use of artificial intelligence.
Now suppose we remove from the equation your reputation, The Family, the controversies, your audience and the storytelling that necessarily accompanies any entrepreneurial trajectory.
Five years from now, what objective fact would allow you to say, “My new way of building companies genuinely works”?
And above all, is there a sufficiently clear outcome that would make you accept the opposite conclusion: “I was wrong”?
Oussama Ammar
Five years from now, I will not ask for this thesis to be judged on view counts, the quality of my speeches or my ability to claim that we are ahead of the curve. The test must be economic and human.
The first objective fact will be the existence of several companies independent of my reputation: products being used, customers who renew, substantial revenue, real margins, teams that attract people better than themselves, and at least a few companies capable of continuing without me. If we build five ventures a year, all twenty-five will not need to become successes for the model to be validated. But the portfolio’s value creation will need to repay the failures many times over, with capital discipline superior to that of a conventional studio or fund.
The second test will concern my use of AI in selection. Across several cohorts, the people identified through this analysis of one hundred answers will have to deliver better results than profiles chosen through my intuition alone or through conventional filters: speed of execution, team retention, quality of decisions, commercial capability and value creation. Without comparison and data, saying that AI helps me recognise potential more effectively would remain a compelling story.
Finally, the system will have to survive its founder. If every venture remains durably dependent on my audience, my energy or my daily involvement, I will not have created a new way of building companies. I will have created a highly sophisticated occupation centred on myself.
What would prove me wrong? Five years of projects kept artificially alive by my reputation; no autonomous leader; a selection process that performs no better than chance or the network; companies unable to retain their customers; and value creation lower than the time and capital invested. At that point, the conclusion would not be that the world was not ready. It would be much simpler: I had a seductive thesis, and it was false.
Background
About Oussama Ammar
An entrepreneur and investor based in Dubai, Oussama Ammar has been the founder and CEO of Naijm since July 2026, a company specialising in embedding artificial intelligence at the heart of business operations. Naijm works on processes with high economic value, with the aim of transforming them into systems capable of operating in production and measuring the gains achieved.
Since 2023, he has also operated as a Solo Capitalist, without a traditional fund structure. He says that he builds around five ventures a year, directly committing his time, capital and responsibility alongside entrepreneurs selected at a very early stage. His portfolio spans artificial intelligence, medicine, media, video games, real estate and business acquisitions.
Oussama Ammar is best known for co-founding The Family, launched in 2013, an organisation that supported several hundred European startups and profoundly influenced French entrepreneurial discourse during the 2010s. He left The Family in 2021.
Between 2018 and 2020, he was also a member of the Global Tech Panel, an initiative launched by the European Union to foster dialogue between technology leaders and public decision-makers on international digital issues.
His current work focuses primarily on venture creation, entrepreneurial selection and the use of artificial intelligence as a tool for analysis and decision-making.