神鱼：AI 时代的知与行

Recently, Cobo co-founder Shen Yu and Extreme Chain Technology founder Jin Ming held an in-depth dialogue, spanning multiple dimensions from investment frameworks, Bitcoin, DeFi, and AI Agents to the meaning of life.

During the conversation, Shen Yu shed the early "mining giant" halo, presenting a highly rational self-reflection. He candidly admitted that his early success involved an element of luck and reviewed his cognitive leap in viewing DeFi as "a condensed 200-year financial history." At a time when AI is deeply taking over cognitive input, this frontier builder steadfastly adheres to long-termism, maintaining sober and uniquely independent thinking amid the impetuous crypto circle.

The following are the core observations and insights distilled from this dialogue:

"DeFi is a condensed course in financial history"

Shen Yu regards DeFi as an experimental laboratory for understanding finance. Mechanisms such as lending, market making, and liquidity mining fundamentally re-answer questions that traditional finance has already addressed. Participating in protocols firsthand is equivalent to experiencing centuries of financial product evolution in an extremely short period. For him, DeFi's value thus became the motivation driving him to make up for his knowledge of financial history and theory.

His investment framework, therefore, grew out of practice. Shen Yu noted that in the early days he had no complete investment framework; the approach he later developed more closely resembles a cycle: practice first, accumulate experience in real markets, then abstract patterns from results, form a framework, and ultimately train it into instinct. The key is to continuously raise questions through action and then fill in financial history and theory in reverse.

"As AI pushes execution costs down, human will becomes even more important"

AI is dramatically shortening the distance from idea to realization. In the past, the main obstacle to many ideas was excessively high execution costs; now, knowledge, tools, and the capacity for action have been compressed into the hands of individuals. As long as there is a sufficiently clear "seed," rapid experimentation becomes possible. The more powerful the technology, the more important one's sense of direction, judgment, and will for sustained action become.

But the proliferation of possibilities also amplifies builders' confusion. AI brings nearly unlimited space for executable projects, making "choosing the right topic" the greatest challenge. Shen Yu admitted that during the mining era he possessed a strong sense of mission and creative impulse, yet now he finds himself in a more confused phase, still searching for a direction large enough and systematic enough. Nevertheless, he still defines himself as a builder: one of the meanings of existence is to add possibilities to the world that did not previously exist.

"Agents are evolving from efficiency tools into an external brain for cognition and self-understanding"

Shen Yu has already deeply embedded AI and Agents into his daily life. On the information input side, the work of reading original long-form articles, screening materials, and compressing them into PPTs is completed by a note-taking system based on evolutionary thinking that he built himself, along with AI-curated daily briefings. For daily interaction and cognitive assistance, he has configured three Agents, each with its own role: one is a tool-rationality Agent that provides highly rational action plans and decision recommendations; one is a softer, more feminine-coded Agent responsible for emotional analysis and empathy, which once astutely pointed out that he "may have death anxiety"; and one is a blind-spot-alerting Agent that, combined with his historical information and personal behavior, specifically helps him identify cognitive blind spots. The role Agents play has already surpassed that of traditional assistants, extending into a companion-like "cognitive external brain" that not only processes information and tasks for people but also deeply participates in emotional care, blind-spot correction, and the construction of a secondary thinking system.

"Core assets must first have a sufficiently large long-term value space"

During the interview, Shen Yu also revealed his first principle for screening core assets: the TAM must be large enough, and there must be an underlying macro vision capable of supporting long-term holding. He believes that only when an asset not only presents short-term opportunities but also has the potential to participate in shaping technological, financial, or social structures over the next decade or even longer does it merit inclusion as a candidate for a core position.

He employs a three-tier position management logic: first, discover with a small position; then, increase allocation through continuous validation; and finally, after seeing the long-term vision clearly, move into the core phase, where the position may be raised to approximately 20%. Currently, the only assets that have truly passed this test are Bitcoin, Ethereum, and Tesla, while SpaceX remains in the observation and timing phase.

Whether Ethereum can remain in the core position depends on its value in the next cycle of infrastructure. Although Ethereum has already entered his list of core assets, its status is being reassessed. The new criterion is whether Ethereum can further become the operating and collaboration infrastructure for AI Agents. If it can absorb the new demands of the machine economy, Ethereum may still maintain its core status; otherwise, the original investment thesis needs to be re-evaluated.

"Bitcoin is better gold"

The essential difference between gold and Bitcoin is a confrontation between two paradigms and camps—namely, the "digital currency + AI" camp versus the "gold + industrial manufacturing power" camp. Shen Yu believes that "Bitcoin is better gold," with the most fundamental reason lying in its complete absence of intermediaries and the asset disposition rights it grants to sovereign individuals. Traditional gold often requires layers of intermediaries and institutional custody to achieve title confirmation and liquidity, whereas Bitcoin can be fully controlled and disposed of autonomously by individuals without relying on any intermediary institution, truly realizing individual asset sovereignty.

There are two types of technology that genuinely and significantly enhance the capabilities of sovereign individuals. Bitcoin changes how individuals control and dispose of assets, allowing value to be held and transferred autonomously, independent of traditional institutions. AI, on the other hand, expands the boundaries of individual capability itself, enabling a single person to accomplish knowledge-based and execution work that previously required an organization. The former strengthens asset sovereignty, while the latter amplifies individual productivity.

"A philosophy of fewer but deeper investments"

Traditional asset allocation emphasizes reducing portfolio volatility through low-correlation assets such as gold, stocks, and Bitcoin, but Shen Yu does not endorse this investment philosophy. "Human cognition and energy are limited, and the assets one can truly understand in depth are actually very few." Rather than purchasing gold one does not understand merely to satisfy asset allocation theory, Shen Yu's investment philosophy is closer to "fewer but deeper," believing that defensive capability does not come from holding more asset classes—because correlations converge under extreme conditions. When a truly systemic crisis (extreme scenario) occurs, those assets you thought would counterbalance one another often end up collapsing in sync or performing identically. Diversification works in extremely stable markets, but it tends to fail in the face of real "disasters."