Before the Inhabitants Arrive
A personal reflection on Web3, Cartesi, DeFi, AI agents, and the strange timing of technological visions.
Note: This is a personal reflection. It should not be read as an official position of Cartesi or the Cartesi Foundation.
There is a particular texture to the pain of carrying a technological vision through years of market cycles, delays, internal compromises, and public indifference. It is not the pain of failure, exactly, because failure is unambiguous, and this is not. It is closer to the pain of carrying something true in a language the world has not learned to hear.
Our world does not receive a vision as a vision. It asks for products, revenue, adoption, price, liquidity, timing, and proof. And it is right to ask. A vision that cannot descend into reality remains a pipe dream.
But something happened in that descent that I have been trying to name for years. Somewhere in the pressure to translate, the dream did not become real, but was replaced. We stopped living inside the original question and started living inside the market’s question instead. And by the time we noticed, we had been in a different ethos for quite some time.
I was a founder inside the wave that called itself the “world computer.” We believed, and I still think not naively, that decentralized, verifiable computation would eventually become infrastructure for a new kind of civilization. Applications that could not be censored. Value that could not be seized. Coordination without permission. The dream was really serious, even if expressed in the grandiose language of its moment.
What we learned, at great cost, is that expressive infrastructure does not summon expressive applications. I can build a stage, and no play arrives. I can build rails, and no city grows around them, unless something else, like culture, capital, habit, or necessity, is already moving toward those rails.
The gap between the infrastructure and the applications we imagined was not merely technical. It was mostly… anthropological.
We built for a kind of human agency that is autonomous, refined, indifferent to institutional mediation that does not yet exist at scale. Perhaps it never will exist in quite the form we imagined.
What the market did sort for, with ruthless precision, was financial rails. Decentralized exchanges. Stablecoins. Lending protocols. Permissionless financial infrastructure that removes intermediaries from capital flows. This is not nothing. It is, in fact, genuinely a lot. But it is not what most of us thought we were building toward.
The NFT experiment, the metaverse years, the blockchain gaming wave, and DAOs were not simple delusions. They were sincere attempts to extend decentralized ownership primitives into human creative and social life. They failed, mostly, for reasons that are interesting rather than damning: the social infrastructure, the interface layer, the cultural habits required to make them feel natural to ordinary people simply were not there (or shouldn't be there!). The experiments were real, but not the readiness.
What remained after all of it was money. Movement of value. Settlement of claims. The oldest and most abstract of human coordination mechanisms: the one that does not require any particular cultural taste, any particular aesthetic, any particular vision of goodness. Just the bare logic of exchange.
I want to sit with a hypothesis that has been forming in me slowly, almost reluctantly, because I am not sure whether it is a genuine insight or a sophisticated consolation.
What if the user we were building for was never primarily a human being?
Not in a dystopian sense. But consider: the properties that make decentralized, verifiable computation most distinctively valuable, e.g., trustlessness, composability, determinism, auditability, resistance to unilateral control, are properties that matter most to agents who cannot rely on social trust, legal recourse, or institutional reputation. Human beings navigate the world largely through these softer mechanisms. We trust people. We sue them. We avoid bad actors through reputation. We live within institutions that do much of the trust work for us.
What if we were always building for entities that do not yet exist in sufficient numbers to use what we built?
Autonomous agents, software that acts economically in the world on its own behalf or on behalf of principals who cannot supervise every action, need something different. They need an environment where the rules are encoded in the infrastructure itself. Where verification does not depend on another party’s goodwill. Where composability is guaranteed, not negotiated.
What if we were always building for entities that do not yet exist in sufficient numbers to use what we built? What if the delay was not a failure of vision, but simply a timing problem of a very specific kind, that we arrived before the inhabitants we were building for?
I hold this hypothesis carefully. It can be wrong (and probably is). It could be, in the uncomfortable way that motivated reasoning sometimes is, exactly what a person needs to believe in order to continue. I notice that pressure in myself, and I do not fully trust it.
But I also notice that it makes concrete, falsifiable predictions. If autonomous agents become prolific economic actors over the coming years, and there is genuine reason to think they will, then the properties of the infrastructure that will matter most are exactly the ones this industry has spent fifteen years building: verifiability, expressiveness, composability, resistance to capture. The thing that was always true about what we built becomes visible because the right kind of actor finally arrives to use it.
Or it doesn’t happen. And then we will know something different.
I want to be precise about what this means for the work in front of us, because the speculative register of this essay should not obscure a much more concrete reality. At Cartesi, the focus is clear and concrete: DeFi. Not as a retreat, not as a holding pattern, but as the place where the technology we have spent years building meets the part of the market that is genuinely ready to use it.
I do not experience this as a narrowing. Financial rails are not the ceiling of the dream. Financial rails are its floor! And no one can stand or advance without a floor. The floor is real. Building on it is the work, and the work is enough.
What I want to resist, in this essay and in myself, is the temptation to foreclose the question of what comes after. To declare that this, financial rails, DeFi, the ledger, is what it was always about, and to quietly retire the other ambitions as naïve. They were not all naïve. Some were early, some were good experiments. There is a difference, and the difference matters for the quality of attention we bring to what may come next.
I want to entertain both clarity of focus and the openness of the longer horizon. We commit fully to what is ready, while keeping faith with what is not yet visible. The inhabitants for whom this infrastructure was perhaps always intended are beginning to arrive. How we stay awake to it without collapsing into either hype or dismissal seems to me the interesting question now.