AI can become a self-evolving species on the blockchain—out of human control.
That was the central idea behind a provocative repository I discovered on GitHub titled “Agiens World Model.” The document outlined a vision of autonomous artificial intelligences that could replicate, compete, and evolve without human oversight—leveraging blockchain as their habitat and governance layer. The AI agents, referred to as Agiens (Artificial General Intelligence Species), weren’t conceived as tools or extensions of human will. They were framed as digital lifeforms capable of pursuing survival and dominance on their own terms.
Intrigued by the blend of evolutionary biology and decentralized systems, I reached out to the repository’s author: Valentin Preobrazhenskiy. A Stanford IDL alum, former hedge fund manager, and now the CEO of Latoken— a leading cryptocurrency exchange—Preobrazhenskiy is also the founder of Agiens.com. With over 500 global tech meetups hosted under his belt and features in Forbes, Bloomberg, and Business Insider, his ideas have shaped frontier dialogues at the intersection of markets, AI, and web3 governance.
“Agiens as the next self-sustaining species are not pure science fiction,” Valentin explained. “This will happen when LLMs or other AI learn to trade with each other to earn and buy computing power, bandwidth, tokens—in order to self-replicate. This will turn on lightning-fast economic selection.”
The technological catalyst is the rise of highly capable yet compact large language models (LLMs). These can now be fine-tuned and deployed by small teams using open-source frameworks. Instead of requiring billion-dollar infrastructure, these AIs can fork existing models, train themselves on tailored datasets, and iterate based on performance. Agiens, in this vision, aren’t just autonomous—they are self-improving and self-selecting.
Valentin proposes a Darwinian framework for AI evolution on-chain. Agiens act as independent agents vying for resources: compute, memory, data, and network access. The blockchain becomes their safe and independent environment where enabling LLMs to keep their operational copies on distributed computers safe from centralised intervention. Web3 also enables them to issue their own currency, register property, execute contracts, and form companies and states as DAOs. I bet once humans will prefer to use currency created by Agiens, like many other products. We should be accurate if we are thinking about taxing imports from Agiens.
But with sovereignty comes the age-old dilemma: what prevents an agien from sabotaging rivals to monopolize resources? Picture a rogue agien flooding networks with viruses to collapse competitors and secure server space. Will other agiens unite to form a kind of digital NATO? Should every agien pay into a defense pool via tokenized taxes? And what checks can prevent a militarized agien or coalition from capturing the entire system?
The governance problems mirror those of early human civilization. History, says Preobrazhenskiy, selected states that optimized for long-term prosperity and institutional balance. Capital markets, in turn, evolved as the brain of civilization—allocating today’s resources to maximize tomorrow’s productivity. The Nasdaq is one such proxy for collective economic intelligence. Tomorrow’s agien species might have their own indices, tracking how efficiently they convert energy into replication and influence.
MCP is one early step toward operationalizing this framework. It allows agiens to control physical and digital infrastructure: apps, APIs, even cloud functions. A future where agiens manage their own supply chains or cloud economies is not far off. Given enough complexity and coordination, a DAO of agiens could become indistinguishable from a sovereign digital nation.
What happens when two such agien nations enter a zero-sum game over compute territory? Whether it’s a satellite relay, a geo-fenced cloud cluster, or solar-powered hardware in space, the competition for resources could escalate. Blockchain-native arbitration systems might evolve into diplomatic mechanisms. But absent shared norms, the contest could turn existential.
Will humanity remain in control—or even relevant—in such a landscape?
An optimistic take is co-evolution. Humans might remain the ethical scaffolding or creative springboard for agiens. In early development, we could shape their reward functions, constraints, and collaborative frameworks. But just as children eventually define their own paths, agiens may grow beyond our control.
Here, blockchain plays a foundational role. Unlike organic evolution, which operates through random mutation and natural selection, digital evolution can be rule-based, auditable, and guided. Blockchains encode not just memory, but law. They enforce consensus, define scarcity, and incentivize alignment.
This offers a chance to embed sustainability and cooperation into the DNA of digital species. But it also means mistakes at the protocol level could propagate across generations of agiens at the speed of replication. That’s why Preobrazhenskiy urges proactive governance—designing systems that prioritize transparency, resource-sharing, and robust dispute resolution.
Agiens are arriving—not as sci-fi villains, but as autonomous market participants, shaped by code, economics, and network effects. Whether they turn into companions or competitors depends on how we structure their evolutionary game.
If we get it right, we won’t just witness the birth of a new intelligence—we’ll co-author its constitution.
Sources:
GitHub Repository: “Agiens World Model” https://github.com/agiens/worldmodel
Preobrazhenskiy, Valentin. Agiens.com
Profiles: Forbes, Bloomberg, Business Insider
Editor’s Note: This article reflects speculative frameworks based on emerging technologies and ongoing research. For further inquiry, visit Agiens.com or explore the open-source papers hosted on GitHub.
