AI Agents and Capitalism: Reimagining The Fundamentals
Imagine a world where your highest-performing colleague isn't human. A world where strategic decisions, financial analyses, and even client interactions are managed by digital entities working tirelessly, 24/7, never asking for a raise. This world isn't science fiction anymore — we're entering it at full speed.
In 2025, as AI companies' valuations surpass the $3 trillion mark and over 40% of administrative tasks are already automatable according to McKinsey, early-stage investors face an existential question: can AI agents — these emerging digital workers — morally coexist with the very foundations of capitalism as we know it?
"Technology is never neutral," wrote Melvin Kranzberg. "It's neither good nor bad, nor is it neutral." This ambivalence perfectly characterizes the silent revolution unfolding before our eyes. To decipher its profound implications, let's return to the fundamentals and dare to reimagine the future together.
1. The essence of capitalism: a dance between capital and production
Capitalism can be defined as an encounter, a confrontation between capital and labor — this age-old dance where one cannot exist without the other, yet where the balance of power has never been equal. In this system, every organization becomes the material expression of capital — and thus its holders — orchestrating human labor to grow its own existence.
As Thomas Piketty observed in Capital in the Twenty-First Century, "the past devours the future": accumulated capital naturally tends to dominate the value produced by labor. The numbers confirm this: since 1978, American workers' productivity has increased by nearly 150% while their real compensation has grown by only 18%, according to the Economic Policy Institute.
In this economic theater, capital drives the movement while labor accelerates it. Wealth springs from capital like a river from its source, with labor providing only the means of existence — like water wheels capturing energy without ever possessing the water that animates them. Wealth doesn't inherently come from labor, which must enable workers to optimize their performance to maximize the creation of capitalistic value.
"Labor is the substance and immanent measure of values, but it has no value itself," wrote Marx. This observation takes on a new dimension today: labor represents a cost for capital, which must generate a return — this cost becoming a calculated investment for future value creation, an equation that algorithms now master better than anyone.
2. The rise of AI agents: the new gladiators of the economic arena
AI agents emerge as autonomous digital entities infused with Artificial Intelligence, capable of perceiving their environment with superhuman acuity, making decisions based on billions of variables, and executing actions to achieve predefined objectives with surgical precision.
"Artificial intelligence is the new electricity," proclaimed Andrew Ng. If electricity transformed every industry a century ago, autonomous AI is now metamorphosing the very nature of work. According to a Gartner study, by 2027, over 80% of companies will exploit commercially viable AI agents, compared to less than 5% in 2023.
Essentially, these AI agents are "digital employees" — but employees who never sleep, never take leave, and whose learning curve defies all laws of human pedagogy. Their architecture revolves around four fundamental dimensions, mimicking with disturbing similarity the integration of an employee in a company:
A. Perception: artificial senses
This first dimension gives the agent the ability to perceive its environment with precision that our biological senses cannot match. Through a network of integrations (APIs) and recognition models, the agent captures reality as structured data. Like a new marketing manager presented with the company's tools, organization chart, and internal processes, the agent absorbs in milliseconds what a human would assimilate in several weeks.
B. Reasoning: probabilistic intelligence
This crucial dimension relies on Large Language Models (LLMs) or, emerging trend, Small Language Models (SLMs) that are gradually imposing themselves due to their reduced energy footprint — a modern SLM consuming up to 25 times less energy than an equivalent LLM for certain tasks. These sophisticated neural architectures, trained on corpora sometimes exceeding 10^15 tokens of data, represent probabilistic models capable of inferring optimal decisions in the face of uncertainty.
This process resembles the acclimatization of a newly recruited financial analyst: they would be given the transaction history, past portfolio performances, sectoral analyses — a baggage they would assimilate through the prism of their training and experience. The difference? The AI agent simultaneously integrates the equivalent of several decades of professional experience.
C. Execution: performance without fatigue
In this third dimension, the agent transcends simple analysis to embody pure action. It performs concrete tasks with mechanical rigor: responding to hundreds of emails simultaneously, optimizing a global supply chain in real time, or orchestrating thousands of high-frequency trading orders in microseconds. A Deloitte study reveals that AI agents deployed in customer service reduce response times by 85% while improving customer satisfaction by 37%.
The agent thus performs work in an optimized, probabilistic manner infinitely faster than a human employee — without ever experiencing fatigue, doubt, or distraction.
D. The feedback loop: perpetual learning
In this final dimension that closes the cycle, the agent's work is adjusted through structured and quantified feedback. It's "coached" not through office conversations or quarterly evaluations, but through precise mathematical adjustments to its parameters and objective functions. This continuous reinforcement process allows the agent to skill up at a dizzying speed, comparable to guiding a junior to become a director — but compressing five years of professional evolution into a few weeks.
This functional tetralogy eloquently illustrates the capacity of AI agents to embody "super-employees" — superior not in their humanity but in their encyclopedic knowledge, superconducting execution speed, and paradoxically, in their very absence of emotions. "The greatest human weakness," noted Friedrich Nietzsche, "is that even our best qualities function intermittently." AI agents, excelling permanently, redefine the very notion of performance.
3. The great reconfiguration: when AI agents reinvent the labor chessboard
Let us imagine for a moment the cold and implacable calculation already taking place in the mind of every business leader confronted with this silent revolution. On one side, a human employee: creative but inconsistent, intelligent but emotional, sometimes loyal but always demanding. On the other, an AI agent: tireless, precise to the microsecond, constantly evolving toward algorithmic perfection, and whose marginal cost tends asymptotically toward zero after the initial investment.
The choice, viewed through the lens of pure economic rationality, is almost a mathematical certainty. As Warren Buffett noted with his characteristic lucidity: "Price is what you pay, value is what you receive." And the value of an AI agent, in purely accounting terms, now eclipses that of a multitude of human workers in a growing number of fields.
AI agents will therefore inevitably disrupt not simply certain professions, but the very concept of work as we have known it since the industrial revolution. This metamorphosis is already visible in the most influential incubators, like Y Combinator — a true barometer of Silicon Valley — where startups fiercely compete to maximize their revenue-to-social-charges ratio by leveraging automation and AI agents capable of handling the majority of repetitive tasks.
"Software is eating the world," prophesied Marc Andreessen in 2011. In 2025, we might paraphrase: "AI agents are eating labor." We are witnessing the galloping automation of tasks once considered inalienably human, regardless of their strategic importance. Legal drafting, financial analysis, medical diagnosis, strategy consulting — all bastions of high-value human expertise gradually succumbing to the assaults of AI agents with stunning performance.
This phenomenon reveals a striking irony: Harvard Business School graduates, long considered the elite of global management, are primarily valued for their case-method training — a methodology that, in essence, makes them organic "probabilistic machines," capable of analyzing complex situations and deducing optimal actions. Don't you recognize there the very functioning of a Large Language Model? Similarly, isn't a successful Venture Capital investor, ultimately, a biological machine for calculating probabilities, evaluating risks, and projecting growth trajectories?
This spectacular substitution raises an existential question: in a world where labor has traditionally fed capital and constituted the functional backbone of our society, how can we maintain social balance if machines progressively but inexorably replace human labor?
How can we prevent our entire planet from becoming the macroscopic reflection of the already glaring inequalities in Silicon Valley — this laboratory of the future where vertiginous technological opulence coexists with stunning poverty, where millionaire engineers daily cross paths with psychotropic-using homeless people, where median household income reaches $138,000 while 15% of the population lives below the poverty line? How can we avert this destiny if work, this fundamental social currency, is captured by non-human entities, and the ex-worker finds themselves orphaned of marketable skills?
For the probability emerges, with almost mathematical clarity, that with the global decrease in economically viable human labor, the main residual positions will be concentrated in the hands of a technical elite capable of developing, refining, and orchestrating these SLMs and designing increasingly sophisticated feedback loops.
In other words, the crucial skills in this world of accelerated automation will no longer be faithful execution or even efficient management, but critical thinking capable of transcending established models and creativity allowing exploration of virgin conceptual territories — these properly human faculties that enable navigation beyond predefined frameworks that even the most sophisticated AI can only reproduce or extrapolate, never truly invent ex nihilo.
Yet — and here the paradox becomes tragic — these very skills that could save humanity from professional obsolescence are becoming scarcer as our civilization advances. We are witnessing a progressive atrophy of critical thinking under the combined effect of labor market hyper-hierarchization that values obedience over innovation, the omnipresence of social networks fragmenting attention, and cultural resistance to change particularly prevalent in Latin societies.
A 25-year longitudinal study from the University of Cambridge shows that deep concentration and systemic thinking capabilities have decreased by 31% among young adults, directly correlating with increased time spent on digital platforms.
As for authentic creativity, this cognitive alchemy born from the fertile encounter of disparate intellectual universes and enriching confrontation with otherness, it erodes inexorably before our eyes with the generalization of remote work that isolates, recommendation algorithms that lock us in comfortable cognitive bubbles, the rise of political extremism that demonizes difference, and the growing deresponsibilization induced by the excessive sophistication and opacity of our socio-technical systems.
"Everything that is not cultivated atrophies," wrote Goethe. Never has this truth seemed more prophetic than today, as we observe the erosion of the very faculties that could constitute our last bulwark against the massive obsolescence of human labor.
4. The dawn of a new paradigm: reimagining civilization in the age of AI agents
These tectonic metamorphoses should not evoke fear or withdrawal — sterile emotions in the face of the inevitable — but rather summon our collective courage to reimagine a civilization in full mutation. This transition, comparable in its scope to the industrial revolution or the advent of agriculture, calls us to confront fundamental questions that transcend traditional conceptual frameworks:
What economic and social model can we conceive for a society where work, this activity that once occupied the majority of waking hours for most humans, is no longer the privileged tool for creating value or structuring existence?
How can we preserve the benefits of a capitalist system — innovation, efficiency, material abundance — without sacrificing on the altar of algorithmic optimization the millions of individuals whose skills are becoming progressively obsolete? Nobel laureate in economics Joseph Stiglitz reminds us that "the market is a good servant but a bad master" — wisdom that takes on a new dimension in the face of AI agents.
How can we systematically cultivate structured contemplation and responsible creativity — these properly human faculties — to guide the performance of AI agents and inspire tomorrow's leaders? Yuval Noah Harari warns us: "Humans risk being transformed into data, while artificial intelligence will become something that resembles the human soul."
These existential questions are not merely academic; they dictate the orientation of our civilizational trajectory. They suggest that it is imperative to fundamentally rearchitect our global higher education model, currently designed to produce qualified executants in a world that now demands critical thinkers and visionary creators.
A longitudinal OECD study reveals that only 17% of current graduates master advanced critical thinking skills, while projections suggest that 65% of jobs in 2035 will primarily require this faculty. We must therefore maximize the development of deep philosophical thinking, intimate understanding of AI's different layers, and the means to keep these prodigious technologies in service of human flourishing — not the reverse.
The worst scenario, of which we already perceive worrying signs, would be to let autonomous AIs take the same insidiously invasive role that recommendation algorithms already exercise in our digital lives — these influence architectures that, under the guise of personalization, progressively reduce our field of possibilities and shape our desires without our knowledge. Such an evolution could only further enslave humanity to the profit of a techno-capitalist elite, inexorably leading to a collective civilizational regression of which history already provides many examples.
For history, that implacable mistress, consistently teaches us that social inequalities pushed to extremes have never engendered anything but unstable, violent, and ultimately self-destructive societies. From the late Roman Empire to pre-revolutionary European monarchies, to contemporary South American oligarchies, the pattern repeats with almost mathematical regularity: excessive concentration of wealth, erosion of social mobility, collapse of collective cohesion, then social cataclysm.
Paradoxically, these technological mutations bring us back, through a strange spiral effect of history, toward a model increasingly reminiscent of the ancient Greek world — particularly classical Athens — where true social value resided in thought, philosophy, and civic participation rather than in repetitive productive tasks, the latter being largely delegated to the slaves of the time. Could this vision, stripped of its morally reprehensible aspects, inspire a better future?
A future where the AI agent would be the modern equivalent of the ancient slave (without the ethically condemnable dimension), where the principal builders would be system developers and designers, where critical thinking and personal fulfillment would constitute the new social hegemony, and where the state, beyond its traditional regalian functions, would have as its primary mission the systematic reduction of inequalities resulting from massive automation.
"Technology should liberate us, not enslave us," proclaimed Steve Jobs. This vision is achievable only on condition of radically transforming our relationship to capital and labor.
For here is the Gordian knot of our dilemma: such a world would require drastically reducing, even eliminating, the domination of capital over human societies — which is by nature antithetical to the traditional role of the investor, historically designed to shape a world serving their clients (the capital holders) who mechanically want to increase their wealth.
This demanding transformation thus implies fundamentally changing the measuring instrument of our civilization: moving from the key indicator "money" to the key indicator "thought" — a Copernican revolution of our value systems that might seem utopian, but is progressively becoming an existential necessity in the face of widespread automation.
This paradigmatic shift would revolutionize the very raison d'être of investors, who would no longer be optimization machines exclusively serving capital, but visionary catalysts serving societal development — whose primary quality would no longer be the capacity for value extraction but critical thinking allowing the envisioning and shaping of desirable futures.
Precursor models are already emerging at the periphery of this vision: philanthropic foundations backed by investment funds, Public Benefit Corporations, collective interest mechanisms in technological advances. Bill Gates' Breakthrough Energy fund, with its $2 billion invested in climate technologies while accepting returns below market standards, offers an embryonic example of this new approach.Another example is Ring Capital’s foundation, funded by a share of the fund’s carry, which is dedicated to supporting non-profits, following a similar path to Raise Capital’s foundation.
"The true tragedy," wrote Oscar Wilde, "is not failing to reach your goals, it's not having any." The objective we must collectively set ourselves is nothing less than the reinvention of a transformed capitalism, where AI agents become instruments of shared prosperity rather than vectors of extreme wealth concentration.
The choice before us, as investors, entrepreneurs, and citizens of the 21st century, is not between accepting or rejecting the advent of AI agents — this revolution is already underway. The real choice, infinitely more subtle and consequential, concerns the vision of civilization we wish to materialize through these prodigious technologies: a new Age of Enlightenment augmented by Artificial Intelligence, or a new Dark Age algorithmically optimized.
The answer to this existential question will determine not only the success of the investments we make today but, more fundamentally, the world we will bequeath to future generations.
Previous articles you'll love :
Catch you soon!
Wonderful post and a very enjoyable and enlightening read. Thanks for sharing
Great article! 👏