Cobo co-founder Shenyu, speaking in a conversation with Jilian Technology founder Jin Ming, outlined how he now thinks about investing, Bitcoin, DeFi, AI agents and personal purpose. The discussion showed a more reflective side of Shenyu, who said his early success included an element of luck and described his later framework as something built out of repeated practice rather than a ready-made theory.
DeFi as a compressed course in financial history
Shenyu said he sees DeFi as a practical lab for understanding finance. Lending, market making and liquidity mining, in his view, are all new attempts to answer questions that traditional finance has already spent generations working through. Participating directly in those protocols lets a person move through what feels like centuries of financial product evolution in a much shorter span.
That experience, he said, pushed him to study financial history and theory more seriously. His investment framework did not begin as a complete system. Instead, it grew out of action. He described the process as a loop: enter the real market first, build experience through practice, extract patterns from outcomes, form a framework and eventually train that framework into intuition. The key, in his account, is to keep generating questions through action and then circle back to fill the gaps with theory and history.
AI lowers execution costs, but makes human will matter more
Shenyu argued that AI is compressing the distance between an idea and execution. In the past, many ideas stalled because execution was too expensive. Now, knowledge, tools and operational capability are increasingly available to individuals, and once a person has a clear enough seed idea, it can be tested quickly.
That shift does not make human agency less important, he said. It does the opposite. As technology gets stronger, a person's sense of direction, judgment and willingness to keep acting become more valuable. At the same time, the expansion of what can be built creates more confusion for builders. If AI opens up an almost unlimited range of executable possibilities, then choosing what to work on becomes the hardest part.
Shenyu said that during the mining era he had a strong sense of mission and an intense urge to create. Now he feels more uncertain and has not yet found a direction that is large enough and systematic enough. Even so, he still sees himself as a builder. One meaning of existence, as he put it, is to add possibilities to the world that did not exist before.
Agents are moving beyond productivity tools
He also said AI and agents are already deeply embedded in his daily life. On the input side, the work of reading long original texts, filtering materials and compressing them into PPT decks is handled by a note-taking system he built around evolutionary thinking, together with an AI-driven daily filtering process.
For interaction and cognitive assistance, he said he currently uses three agents with distinct roles:
- a tool-rationality agent that offers highly rational action plans and decision suggestions;
- a softer, female-coded agent focused on emotional analysis and empathy, which at one point told him he might have death anxiety;
- a blind-spot agent that uses his historical information and behavioral patterns to identify cognitive blind spots.
In his description, these agents no longer function as ordinary assistants. They have become a form of companion-like external cognition, helping with information processing and tasks while also taking part in emotional support, blind-spot correction and the construction of a second thinking system.
The first test for a core asset is long-term value space
When he discussed core holdings, Shenyu said his first principle is that the total addressable market, or TAM, must be large enough and backed by a macro vision that can justify holding over the long run. For an asset to qualify, it cannot rely only on near-term opportunity. It must also have a plausible role in shaping the technological, financial or social structure of the next 10 years or longer.
He uses a three-layer position management approach. The first stage is discovery with a small position. The second is increasing allocation after continued validation. The third begins only once the long-term vision is sufficiently clear, at which point the position may rise to around 20%.
By that standard, he said only Bitcoin, Ethereum and Tesla have truly passed the test so far. SpaceX is still in the observation and timing stage.
Ethereum remains on his core-asset list, but he said its place is being reassessed. The new test is whether Ethereum can become infrastructure for AI agents to run and coordinate. If it can absorb demand from a machine economy, it may retain its core status. If not, the original investment logic would need to be revisited.
Why he calls Bitcoin a better form of gold
On Bitcoin versus gold, Shenyu framed the difference as a contest between two paradigms and two camps: one built around digital currency plus AI, the other around gold plus industrial manufacturing power.
His conclusion was direct: Bitcoin is a better form of gold. The basic reason, he said, is complete disintermediation and the fact that Bitcoin gives sovereign individuals direct control over asset disposal. Traditional gold usually depends on layers of intermediaries and institutional custody for ownership confirmation and liquidity. Bitcoin, by contrast, can be fully controlled and moved by the individual without relying on any intermediary institution, which in his view makes genuine personal asset sovereignty possible.
He added that two technologies have materially expanded the power of sovereign individuals. BTC changes how people control and dispose of assets, allowing value to be self-held and transferred outside traditional institutions. AI expands the boundary of personal capability, letting one person complete knowledge and execution work that previously required an organization. One strengthens asset sovereignty. The other amplifies individual productivity.
A philosophy of fewer assets, understood more deeply
Shenyu also pushed back on the traditional allocation model that uses low-correlation assets such as gold, equities and Bitcoin to reduce portfolio volatility. He said human cognition and attention are limited, which means the number of assets anyone can truly understand in depth is small.
Rather than buying assets he does not understand just to satisfy allocation theory, he prefers a "fewer but deeper" approach. In his view, defensive strength does not come from owning more categories of assets, because correlations tend to converge in extreme conditions. When a systemic crisis arrives, assets that appear to offset one another can end up falling together or behaving in similar ways. Diversification works in calm markets, he said, but may fail in a genuine disaster scenario.