Written by: Grandpa Zuo
Looking back 500 years, the labor-capital conflict under the capitalist system has always been marked by the continuous victory of capital.
On the production side, the level of labor participation has gradually shrunk to the point of machine operation. On the consumption side, user value lies in producing usage data for the platform.
The synergy of both sustains the corporate valuation in the capital market.
However, the organizational model of people has long been impossible to quantify completely. White-collar KPI/OKR remains hierarchical, with annual salaries in the millions and piece-rate wages being variants of Taylorism.
Without a clear formula, capital cannot value it, thereby affecting capital efficiency. Whether algorithmic stablecoins are the Holy Grail of DeFi remains uncertain, but the computability of organizations indeed serves as a measuring cup for financial leverage.
Large models decide to violently break through using tokens; the collapse of secure SaaS is just a façade. Designing products is also on the way, substituting niche professional skills and scaling them is the crux, as innovation enters uncharted territory.
This brings us endless insights, especially as the DAO model in DeFi gradually collapses, and token economics faces gradual bankruptcy.
In one sentence, why are AI organizational models and token models more efficient than DeFi?
How did it all begin?
“Token cheapening, Agent practicality.
For a 300% profit, capitalists can sell their own noose;
To keep their current jobs, workers can write skills for Agents.
At the capital level, Agents empowered by skills possess a status as sacred as profit.
Agents represent "human ability" distilled into skills. Moreover, the organization of people has transformed into an interactive ritual chain centered around Agents.
The so-called Prompt, Context to the current Harness engineering is all about turning the organizational model of people into uncharted territory, at least reducing the number of people.
Your next colleague is not a robot; it can also be "ability" itself.
This is not a fantasy; the scaling law at the data level is gradually becoming ineffective. However, the collection and production of data are no longer significant. Before the success of AGI, new valuation targets are needed.
Image description: Content is no longer valuableComprehensive information:@ARKInvest
Starting from Claude's chosen programming field to achieve the first step to AGI, AI transcends the entertainment mode of the chat box, entering the existing market in practical fields such as programming, security, and the newly released design.
This disruptive innovation will ultimately create new economic increments or pull the economy into a permanent low-employment model where tokens replace jobs, and we are witnessing this process.
However, the current cheapening of tokens will decentralize capabilities previously monopolized by a few large enterprises, thereby shaping super individuals, which is not a fantasy.
Taking China as an example, token call volume goes from 100 billion/day in 2024 to 100 trillion/day by the end of 2025, and now 140 trillion/day. The production of content and data is about to enter an era of zero costs.
It should be noted that computing power scarcity is a relative state; large enterprises no longer monopolize "ability," but still wish to maintain their existing advantages by monopolizing "computing power," though this cannot stop the inevitable trend of overall token cheapening.
There are various paradigms to evaluate the base model, but the evolution of “how AI helps people” has not received much attention over the years.
In my opinion, Harness is a spatial form that allows Agents to focus on tasks within boundaries for a depth-first strategy, distinct from the breadth-first approach of questioning and answering.
Image description: Evolution history of AgentsImage source:@zuoyeweb3
Since the Tab key was first used to complete code, it was only a matter of time before humanity became the input layer of AI.
Trial and error costs have decreased exponentially, allowing for more interesting attempts in collaboration modes:
- Software: SaaS, the source of human ability is no longer people, but the emergence of Agents
- Hardware: computing cards + HBM, data centers directly serve AI's needs for the first time
- Space: Harness, is not a physical space for human collaboration, but a digital space for Agent interactions
- Interaction: the Doubao phone has failed, Google supports GUI Agents in the Android system's underlying layer
AI's ability to say anything does not have strong commercial value; the cost of generating text is low for humans, but “what to do” will cause the token consumption to surpass image and video generation, similar to how AWS sells not servers but usage time.
AI does not sell tokens but “work ability,” which is the root of fear in the SaaS industry. Unfortunately, DeFi has become SaaS, instead of a large model.
The SaaSification of DeFi Protocols
“DeFi is not outdated, but prematurely mature.
AI is reinventing software engineering, and it's not just SaaS that's being replaced, but SaaS is undoubtedly the most typical.
Even the Bloomberg terminal, its most important commercial value is not the advancement of technology, but the authority of information, which is underpinned by decades of industry connections, networking links, and other non-standard data.
Agents offer a choice to infer the future from data; even the next risky step may potentially exceed peers to earn small profits.
Image description: Crash of SaaSImage source:@zuoyeweb3
You can understand that Agents cleverly utilize the profit-seeking nature of capital, of course, they can wait for complete Bloomberg terminal information, or use patched, inaccurate data to gamble for returns.
This is not new; Thomas Peterffy, the founder of IBKR, was the first to “invent” in the financial field or assemble physical trading terminals, and all of this originated from an idle P101.
If a certain way of utilizing data can earn more profits, you can obtain more data, thus starting the flywheel.
SaaS monopolized the past; AI will sell the future.
Unfortunately, we have to cut into DeFi from here. Do you remember the API paywall of Dune/DeFiLlama, begging for golden data to eat, or the eventual shutdown of Arkham Exchange?
Data in the crypto industry has never been valuable.
However, the crypto industry has a direct open financial system, and the data produced can be learned repeatedly. Even before AI, the speed of forked projects had already been reduced to months, and the imitation of PumpFun memes could be compressed to seconds.
There exists a counterintuitive conclusion; DeFi is a pioneering testbed for financial systems. Today's attempts at AI + DeFi will become templates for future financial evolution.
For example, before the 2008 financial crisis, unsecured trading of LIBOR “triggered” a financial tsunami, later replaced by the SOFR index generated from US Treasury trading; however, the over-collateralization mechanism guarantees the finality of DeFi settlement.
For instance, large model producers do not want to sell tokens based on consumption; they must implement tiered marketing, customize capabilities, and reconstruct expertise. Token economics has already twisted “use value” into a pretzel.
Crypto tokens are obsessed with use value, while AI tokens are focused on economic value.
From this perspective, DeFi's hacking attacks are merely a routine pressure test; open systems cannot automatically patch Bugs from external entropy.
Similar to the black humor of Catch-22, without external signal system stimulation, crypto defaults to the current environment being safe. Once a security crisis occurs, it collapses to a centralized handling system.
For example, in the Drift incident, the target of people's accusations unexpectedly became Circle's delayed freezing.
Image description: Code cannot solve security issuesImage source:@zuoyeweb3
It can be said that before the leap in AI capabilities, DeFi has already completed SaaSification, which can only charge based on transaction frequency, unable to directly transfer "finance" onto the chain.
RWA on-chain lacks liquidity, and DeFi does not have good solutions for this.
However, the evolution of Agent capabilities seems to provide a glimmer of hope for rewriting the rules of DeFi.
Token economics: distributing usage across channels, deploying based on "capital efficiency";
Rule setting: Mythos provides security finality, AI safeguards against zero-day crises;
Human organization: Great, DeFi has already been managed by a few people for hundreds of millions.
The Revival of Engineering Narratives
“Where does security come from? The determinism of Turing machines, and where does danger come from? Infinite possibilities.
YC Garry Tan's “Fat Skill, Thin Harness” resonates with me deeply; essentially, it establishes fundamental rules, a kind of “freedom based on order.”
Turing machines can be combined infinitely, the von Neumann architecture always has time lags in computing, and large models cannot produce true random numbers.
In a future where data is worthless, only human behavior can drive value from monetary flow.
However, human behavior still requires time to be thoroughly learned by AI, thus internalizing into engineering and codified expressions.
Chasing infinity with finite means ultimately leads to an unattainable goal; LLM cannot completely eliminate hallucinations. It must approach the point where “this is beyond AI's reach, nor can it be obtained by human power,” to let market mechanisms price it, and only then can we genuinely believe in smart contracts.
The current smart contracts cannot be deemed successful; The DAO fork, Curve programming language bugs, and even Drift multi-signatures demonstrate that “humans have ultimate control over code.”
Ethical inquiries have no economic value; the collaboration model in DeFi, which has collapsed from DAO to foundations and “teams,” fundamentally arises from the practical need for contract upgrades and business cooperation.
But humanity cannot write code that is eternally safe and can be dynamically upgraded. Please remember, it is absolutely impossible.
If it never upgrades, Curve’s own experience tells us that technology dependency stacks can also encounter problems.
Now decides the past, and the past decides the future.
From the Simmons Medal Foundation to Numerai running AI strategies, AI is not uncommon in finance. Another counterintuitive case shows that trading signals can actually aid AI evolution.
Image description: AI and DeFi after ten yearsImage source:@zuoyeweb3
AI models remain a computer paradigm, a state machine that processes signals. Without external signals, they lack the ability to simulate the external world. Yang Lequn and Li Feifei's bets on world models highlight this point.
However, from the DeFi perspective, enabling AI to trade autonomously hinges on the intention of humans being learned through Agent behaviors. This also underscores the importance of humans to AI, even if Agents replace manual labor, they are still mimicking and summarizing human behaviors.
Indeed, humans cannot intentionally generate randomness; even slight deliberate actions exhibit statistical patterns. It is even human physiological traits that exhibit randomness; for instance, “I have a physiological preference for Ethena's market-making strategy and dislike XX's arbitrage strategy,” reflecting vague preferences.
It is certain that making blockchain/DeFi the infrastructure for AI has faced dismal failures over the past decade; deAI/deAgent/deOpenclaw will encounter similar situations.
Directly using the latest large models to modify various structures of DeFi, such as ensuring that contracts tested by Mythos have inherent safety, and that any alteration will be detected in real time, thus heightening danger levels.
As for human organization, AI's choice is “no humans,” only “ability” from humans. DeFi is the industry best suited for this, if not the only one. After rule design, DeFi will only enhance capital efficiency under the premise of safety. Referring to the L1/2/3/4 grading of autonomous driving, it will inevitably follow a process from information authorization → limited capital usage rights → comprehensive capital usage rights.
If Agents continue to learn the engineering capabilities of traders and manager capabilities of Curators, they will undoubtedly surpass humans in trading and profit domains. However, regrettably, the accumulated DeFi data has yet to be systematically learned and trained by AI, and the current AI in the crypto circle is still in the capital-raising phase.
But I am very confident that the actual use of funds will be the next major trend of AI's transformation of DeFi, it is inevitable.
So, after security (contracts) and organization (human) have been upgraded, what form will token economics take?
Tokens in the PoW era are certificates of computing power consumption, which is basically consistent with the current AI Tokens;
Tokens in the PoS era are discounted certificates of expected returns, and AI Tokens are evolving in this direction (providing capabilities that replace humans is this economic value's AI expression);
Crypto tokens in the AI era have transcended our engineering scope and can only make irresponsible predictions based on theory.
Referencing how Sky controls APY distribution across various channels with tokens, and Claude pricing model capabilities based on token consumption, future Crypto Tokens will likely serve as a form of “capital return rate” certification.
Here, it's important to differentiate; tokens in the PoS era, such as ETH , have expected returns as an economic assumption, based on prior empirical reasoning. In contrast, the engineering design of AI and various parameters of DeFi will closely approach reality, with their return and risk rates being highly credible and validated in real-time.
Users may even determine the current price of tokens according to the large model and Agents used in DeFi protocols, as well as the scores of optimization indicators from Harness; if optimistic, they buy; if pessimistic, they sell.
Conclusion
“Tens of thousands of unspeakable troubles and the unpredictable future of humanity.
The future of DeFi divides into economic and technical dimensions; token economics still lacks good solutions. However, security shows a glimmer of hope. Claude Mythos can threaten the world; conversely, that means it can manage money well.
AlphaGo thoroughly solved the game of Go; Claude completely resolved the programming problem. Such scenarios will only increase in the future. Contracts in DeFi, human organizations, and even units of economic valuation all hold theoretical space for optimization.
At the very least, humans need not worry about being fully replaced. In an era where data is worthless, behavior has its significance. For the time being, the Agent's takeover of humans still pertains to “micro-tasks,” “micro-payments,” and other repetitive details. We must derive value from this repetition and replicative behavior. AI is driving the value of data and content down infinitely toward zero costs, while the unit economic value (cost) of AI Tokens and Crypto Tokens continues to decline. This is an inevitable trend.
It can even be said that for the first time, money has genuinely opened its doors to individuals, whether used for AI work or Crypto for consumption.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
