Let's delve into the specific form of Post-AGI finance, as well as the core competitive advantages and specific evolution path of AIUSD under the core vision of "AI-driven financial intelligence."

Written by: Deep Tide TechFlow

You won't know if it's a mule or a horse until you take it out for a walk.

In the past decade, the cryptocurrency industry has transitioned from "talking big in white papers" to "letting on-chain performance speak for itself," with the market favoring projects that can achieve impressive results in practical applications.

On November 19, 2025, an Agentic AI currency platform designed around real trading from day one announced the completion of nearly ten million dollars in Pre-Seed funding, with supporters including early institutional investors from star projects like Anthropic, Sequoia US Scout Fund, a16z Scout Fund, and leaders from Tesla FSD AI among several top Silicon Valley investment institutions.

This is AIUSD: dedicated to enabling both humans and AI to "talk to money" through natural language, where the system can understand intentions and break down tasks, automatically managing, routing, and settling assets across all mainstream public chains, centralized exchanges, and major stablecoins, achieving true "financial inclusion" through blockchain and pushing global finance towards true "intelligence."

At the time of the funding announcement, Deep Tide TechFlow engaged in an in-depth conversation with AIUSD's two co-founders, Yao Meng and Bill Sun.

When discussing AI's performance in finance and the more niche cryptocurrency industry, the recent trading competition among the world's top six AI models also became a topic of conversation. Regarding this competition, the two guests keenly pointed out the fundamental issue that "AI does not have a complete execution environment" in this competition, and expressed eagerness for AIUSD to join the competition. Yao noted:

AIUSD's Agent does not just provide opinions; it is directly connected to real liquidity, funding fees, cross-venue execution paths, and clearing systems. If this competition allows for real execution in the future, we would be very willing to participate.

The confidence in AIUSD's practical operations comes from multiple aspects. Bill excitedly stated that we are currently in a "rare window" for successfully pushing AIUSD to market:

On one hand, AI is transitioning from "being able to chat" to "being able to execute," and the crypto infrastructure is truly maturing;

On the other hand, the underlying funding fee engine, micro-structure execution engine, and cross-scenario routing system of AIUSD have been running in real operations for over two years, with an annual trading volume of one trillion dollars, core strategies averaging over 20% annualized returns, a Sharpe ratio around 22, and no monthly drawdowns. We have also established risk management, compliance frameworks, and risk control automation, ensuring that this is a system that can truly support capital.

Facing the current momentum of AIUSD, Yao shared a clearer plan:

AIUSD has successfully passed the stage from 0 to 1: 0 is proving that this is technically feasible, and 1 is turning it into an account layer that anyone can use; you speak, and the system routes and settles automatically. Currently, AIUSD is at the turning point from 1 to 10, and in the future, we will focus on delivering the same infrastructure stably to a broader consumer and developer ecosystem.

In this issue, let us follow the insights of two Wall Street/crypto OGs as we explore the specific form of Post-AGI finance, as well as the core competitive advantages and specific evolution path of AIUSD under the core vision of "AI-driven financial intelligence."

AIUSD: A "currency" that is easy for both humans and AI to understand and use

Deep Tide TechFlow: Thank you both for your time. Please start with a self-introduction.

Yao:

Hello everyone, I am Yao, and I am glad to communicate with you all.

In 2011, I entered the world of cryptocurrency while I was still in college. I initially participated in mining and gradually got involved in various blockchain projects and the construction of early exchanges. Over the past decade, I have experienced the industry's changes and developments, from the Mt. Gox claims, the FTX collapse, to the recent October 11 liquidation. For example, during the massive liquidation on May 19, 2021, we were wiped out of a $50 million position overnight, and many investors were liquidated; during last month's October 11 liquidation, our company was likely the first large institution in the world to sell off on a large scale; and during last year's election, I heavily invested in Dogecoin, where at one point, 20% of the Dogecoin holdings in the entire market were from our company's self-funded arbitrage strategy, with annualized returns reaching as high as 200%, which can be considered an astonishing yield in delta-neutral strategies.

Over the years, I have always been involved in trading-related activities, from arbitrage and high-frequency systems to institutional custody and quantitative strategies. Now, I am launching AIUSD to repackage the system capabilities we have accumulated over the years and create a stablecoin infrastructure that both AI and humans can use naturally.

Bill Sun:

Hello everyone, I am Bill (Qingyun) Sun, and I am glad to communicate with you all.

I studied mathematics for both my undergraduate and doctoral degrees at Stanford. When I started my PhD in 2014, deep learning was just gaining momentum, and I had a stubborn thought: I must understand the mathematical structure behind these models and see if they can hold up in the real, complex environment of finance. My research has always focused on two things: the underlying principles of deep learning and the practical application scenarios in finance.

In 2016, I worked at Google Brain on NLP, before the term Transformer was even coined. We conducted reading comprehension and question-answering experiments under the attention framework, modifying model structures while trying different tasks. The work we did at that time was later summarized and named Transformer by the outside world, and I can say I was among the early explorers of this route.

During this phase, I systematically observed several phenomena for the first time: one is that different tasks can actually share a unified attention structure, with generalization capabilities stronger than expected; the other is that as the model and data scale increase, performance tends to improve in an approximately logarithmic manner, which later became known as Scaling Law, discovered through extensive experimentation.

After returning to Stanford, I continued to work with two advisors on the mathematical structure of deep learning. One is David Donoho, a member of the US National Academy of Sciences and a recipient of the Gábor Szegő Prize, a representative figure in high-dimensional statistics and compressed sensing; the other is Stephen Boyd, a member of the US National Academy of Engineering, a prominent scholar in optimization, and the founder and head of BlackRock AI Lab. Both advisors not only focus on theory but also deeply engage in the quantitative finance industry: David Donoho previously conducted research at Renaissance Technologies, while Stephen Boyd built and led the AI Lab at BlackRock. Working with them allowed me to encounter early on what kinds of problems are truly important in the real market and which model characteristics can survive outside of academic papers.

During my PhD, I also worked in Quant Research at Citadel and Point72 Cubist, applying deep learning and reinforcement learning ideas to stock and futures trading. After graduation, I went to Wall Street, where I worked as a fund manager at Millennium WorldQuant, trading global stocks and managing a sizable statistical arbitrage portfolio in US stocks.

I also got involved in crypto relatively early, starting around 2015. At that time, I took Dan Boneh's cryptography course at Stanford and worked on projects with Balaji and Lily Liu at the Stanford Bitcoin Lab, where we researched early designs of Lightning micropayments and Ethereum smart contracts, as well as the privacy mechanisms of Zcash and Monero. I also worked on Bitcoin fee optimization and some early attempts at generating stablecoins through on-chain lending.

Personally, I believe the most promising application of crypto in AI is the agentic economy. Since last year, I have also participated as a co-founder and Chief Scientist in the construction of PIN AI, attempting to establish an agentic economy from the perspective of consumer data autonomy and AI personalized voice + recommendations, transforming human intentions into AI actions.

Over the past decade, I have increasingly felt that: the boundaries between AI, crypto, and traditional finance are rapidly disappearing. Traditional finance brings mature efficiency and risk control systems, crypto reshapes the underlying infrastructure, while AI turns all of this into truly "intelligent" systems. The intersection of these three areas is essentially rewriting the underlying logic of global finance.

AIUSD was born in this context and long-term thinking. I want to create a "currency" that is both natural for people to use and can be inherently understood and invoked by agents, capable of automatically completing routing, settlement, and risk control in a multi-chain, multi-market environment. In simple terms, it is about leveraging this new form of currency to push global finance from "automation" to a truly "intelligent" stage.

Deep Tide TechFlow: Both of you are very early participants in crypto. Could you share why you wanted to create an AI x Crypto project like AIUSD? What problems does it aim to solve?

Yao:

Having been in the space for 14 years, my biggest realization is: while blockchain has created a tremendous wealth effect, it has not truly achieved "financial inclusion."

The barriers to entry in this industry are too high. I don't lack money; I want everyone to use digital currency. If this is my legacy, I would be very happy. So, after more than a decade, the truly active users on-chain are mostly those chasing airdrops and gamblers, rather than application-layer users who can benefit from the technology. I have always considered the question of how to enable ordinary people, like your 50-year-old parents, to use cryptocurrency without barriers, just as they do with WeChat and Taobao.

I believe we should start with wallets to reshape the entire crypto trading experience. But sadly, I have been in the space for 14 years, and there still isn't a multi-chain, one-click trading wallet on the market, which is disheartening. The reason is quite simple: the crypto space primarily has exchanges as the profitable business, so all resources have leaned towards exchanges. Meanwhile, wallets, as the largest entry point, do not have good products.

For example, I want to buy a token on BSC, but my funds are on Solana. I first need to exchange SOL to pay for gas, then exchange BNB to pay for gas on another chain, and I also have to deal with cross-chain transfers, bridging assets, calculating slippage, etc. The entire process requires switching wallets, waiting for confirmations, and handling fees, which is almost "unusable" for ordinary users. Suppose your user is not a PhD from Stanford but a taxi driver; why would he want to deal with such complexity? This makes it very difficult for the product to truly integrate into people's lives.

The emergence of AI has brought a turning point. AI is inherently good at understanding intentions, making decisions, and automating complex processes. AIUSD leverages AI to abstract away all these cumbersome operations, allowing cross-chain transactions, routing, settlement, and profit distribution to be completed automatically.

In other words, the core of AIUSD is a unified wallet layer that allows both AI and human users to perform financial operations with zero gas fees and zero cross-chain barriers.

We believe that AI will be the primary mode of interaction in the next generation: the money of the future will not be "manually clicked" by people, but rather AI will directly understand what you want to do, then automatically find the optimal path, settle, generate returns, and even manage risks. Therefore, the problem AIUSD aims to solve is essentially "making money AI-native." You don't need to understand ten chains or twenty token standards; you just need to tell an AI, "Help me transfer a thousand to a friend," or "Help me buy some BTC and put it into stable returns," and the AI can automatically complete the routing, trading, and settlement.

We often say that AIUSD aims to become "One AI to rule them all": a unified infrastructure that allows everyone and all intelligent agents to naturally use funds.

Not everyone needs to become a blockchain expert; users only need to say a single sentence to let the system handle everything for them. This is the direction we have believed in from day one: abstracting the complex financial system into commands that can be invoked with natural language.

This project is actually a continuation of what we have always been doing: making capital operations programmable, verifiable, and reusable. Previously, it was HFT bots; now it is AI agents. The underlying logic hasn't changed; it has just become smarter and more abstract.

Bill Sun:

I entered the crypto space in 2015, and over the past decade, I have experienced the entire process of the industry evolving from the conceptual phase to the formation of infrastructure. Along the way, I have two very intuitive judgments.

First, I believe stablecoins and RWA (Real World Assets) will be the key entry points for real-world liquidity to truly enter the chain. Regardless of market fluctuations, the demand for these two types of assets is the most stable and closest to real economic activity.

Second, to make on-chain money truly usable, we must completely abstract away the complexity of on-chain operations. For ordinary users and institutions, managing addresses, selecting chains, and changing RPC should not be barriers to usage.

I have always believed that stablecoins will ultimately become First-Class Citizens in the blockchain world. Today, stablecoins are still considered "subjects" under assets like ERC-20, while gas tokens like ETH, SOL, and BNB are regarded as first-class citizens. But in the future, it should be the other way around: users holding USDC or USDT should not need to know which chain it is on, just like when you transfer dollars between Chase, Interactive Brokers, or Robinhood, you never care about how the underlying system operates.

The current multi-chain structure resembles multiple religious states, each independent and closed off, with cross-chain bridges being both complex and high-risk. One of our important goals in creating AIUSD is to abstract away this sense of fragmentation, allowing "using money" on-chain to feel as natural as it does in the real world.

From an AI perspective, I have always felt that the industry lacks a truly machine-native currency. It needs to facilitate micropayments, be precisely executed by APIs or function calls, and ideally allow models to directly generate deterministic DSLs (Domain-Specific Languages) to express the flow of funds in a code-level manner. Because of this, we have merged two things together:

• At the crypto layer, unify multi-chain, multi-pool, and multi-asset into a "money" experience;

• At the AI layer, make money an object that machines can directly understand and invoke.

Ultimately, I hope AIUSD is not just a stablecoin in the M0 form but an intelligent asset system with M2 characteristics: it accrues interest like a money market when not in use, and when needed, it can leverage, hedge, or exchange for spot, encapsulating the core capabilities of traditional tools like Interactive Brokers and money market funds into the backend of the stablecoin. This way, it is friendly to both humans and intelligent agents, naturally becoming the underlying asset of the next generation of financial systems.

Now is a rare window for AIUSD to reach the consumer end

Deep Tide TechFlow: AIUSD has three core keywords: Crypto, AI, and stablecoins. You both are at the forefront of Wall Street; how do traditional institutions represented by Wall Street view the potential of these three tracks?

Yao:

To be honest, Wall Street's attitude towards these three directions is very layered.

The underlying infrastructure of crypto, such as custody, settlement, and stablecoin clearing networks, is no longer seen as "rebellious things," but rather as a new generation of financial pipelines. More and more funds are researching "how to make their capital efficiency closer to on-chain speed."

AI is another layer; Wall Street's quantitative and risk control has already become highly AI-driven, especially in strategy generation, data cleaning, and sentiment recognition. However, everyone has found that while AI can perform analysis, it cannot directly execute financial actions, which is a significant gap.

Stablecoins serve as the bridge connecting the two. They are the energy for AI to execute trades and the unit of settlement in the crypto world. Over the past five years, USDT and USDC have proven the existence value of "on-chain dollars"; in the next five years, the market will need "smarter money" that understands strategies, returns, and risks, and can be orchestrated.

So from the perspective of Wall Street OGs, these three tracks are not independent; they form an evolutionary chain: AI is the demand side, crypto is the infrastructure, and stablecoins are the intermediary layer. What AIUSD aims to do is to integrate these three layers into one.

Bill Sun:

My experiences at WorldQuant, Citadel, Point72 Cubist, and Millennium have made me very familiar with the "compliance-first" culture of traditional buyside. Over the past two years, from Bitcoin and Ethereum ETFs to the continuous advancement of stablecoin legislation (including the so-called Genius Act), a very clear signal has been released: traditional institutions can enter the market.

Once institutions can truly enter the market, they often bring about geometric levels of new liquidity, which is something that can easily be overlooked in the crypto market structure but has far-reaching implications.

In this context, stablecoins and compliant custody systems (like Coinbase Custody; on the offshore side, CEFFU) are transforming the so-called "on-chain dollars" into a trustworthy, regulated settlement layer. I predict that a batch of institutional-level stablecoins will emerge, possibly from large internet companies, payment groups, or even offshore exchanges. Stablecoins will gradually transition from a landscape dominated by two or three players today to a multi-entity competitive phase. This change will make on-chain dollars more like an open settlement network rather than just products from a few companies.

The application of AI in trading is still at the stage of "assisting human trading and research efficiency" on Wall Street as a whole. Before I started my company in 2023, I conducted a small experiment: I focused on AI, fintech, and crypto-related stocks with an AI analysis system for a year, achieving a return of about 880%. Of course, this is a high-volatility strategy, but it validated one thing: AI does deep research, while humans make PM judgments; this combination is far more efficient than I initially imagined.

Because of this experiment, I began to seriously consider whether this capability could be turned into a product that allows ordinary retail investors to access institutional-level research and decision-making tools.

Changes on the regulatory front are also driving another thing, which is the opening up of Tokenized Stocks. US stock trading is moving closer to a true 7×24 model from "partially extended trading hours," and stock trading for US investors will gradually shift from brokerage systems to on-chain. Tokenized stocks will become a compliant connector, linking traditional financial markets with the on-chain world.

Deep Tide TechFlow: The team has been deeply involved in the LLM Trading Agent track for many years and has undergone a long R&D process before officially launching AIUSD. Do you think now is the best time to launch AIUSD? What preparations has the team made to better bring AIUSD to market?

Yao:

I think now is a very good window period, and we have actually prepared for this step for many years.

The strategy engine, trading system, and risk management framework behind AIUSD all stem from the underlying stack we used for high-frequency trading and arbitrage in the past. Over the past two years, we have been running the AIUSD prototype internally, achieving an annual trading volume exceeding one trillion dollars without external exposure. At the same time, we are deeply collaborating with CEFFU in the Binance ecosystem to connect institutional custody and MirrorX yield channels; additionally, we have built a complete risk, compliance, and risk control mechanism, from capital flow monitoring to multi-signature and limit management, all adhering to institutional standards.

This is why we are confident. Now is the time for AIUSD to truly reach the consumer end and the agentic economy.

Bill Sun:

Yes, it is indeed a rare window period.

On one hand, AI is transitioning from "being able to chat" to "being able to execute": more and more agents need to autonomously schedule accounts, place orders, make payments, and settle, but there is still no unified, scalable, and sufficiently secure account and settlement infrastructure in the industry, and AIUSD can fill this gap.

On the other hand, the crypto infrastructure is truly maturing: stablecoin clearing has become standardized, cross-chain routing is gradually becoming reliable, institutional custody systems are improving, and liquidity markets are deepening. In such an environment, we can achieve safety, efficiency, and compliance simultaneously, rather than having to choose one of the three.

We have retranslated the technology stack of quantitative trading and risk management over the past decade into a zero-threshold account layer. As Yao mentioned, our underlying funding fee engine, micro-structure execution engine, and cross-scenario routing system have been running in real operations for over two years, with an annual trading volume of one trillion dollars. We have also established risk management, compliance frameworks, and risk control automation, ensuring that this is a system that can truly support capital, rather than just a demo.

In summary, we are not simply "hanging a model on an exchange," but rather encapsulating "exchange-level execution and clearing capabilities" into an account layer that both humans and AI can directly use.

Designing Around Real Trading: AI-Driven "Intention → Strategy → Execution" Closed Loop

Deep Tide TechFlow: As the first core product aimed at C-end users, what differentiated advantages does AIUSD have compared to other AI trading platform products? What do you think is the most attractive feature of AIUSD for C-end users?

Yao:

The biggest difference is that AIUSD is not a strategy platform, but an account layer.

Many so-called AI trading platforms merely allow users to select models, choose strategies, and place orders, but the underlying mechanism is still centralized matching, with funds isolated, and AI merely acting as a advisor.

AIUSD is the opposite; we are creating an "account abstraction + intelligent settlement" layer. Users only need to hold one AIUSD, and regardless of whether it's spot, contracts, staking, or cross-chain payments, all liquidity and yield routing are completed automatically.

AIUSD is not about creating a "smarter trading platform," but about enabling all platforms to settle and flow more intelligently through it.

Bill Sun:

Yes, we can illustrate what we are doing with a few specific examples.

First, in AIUSD, natural language is the instruction: Users no longer need to click buttons, switch chains, or calculate gas. For example, a user can simply say, "Use 1000 AIUSD to buy some ETH, and put the rest into stable returns," and the system can automatically break this down into a series of trading, routing, and settlement actions, executing them in the background. For users, it's just one sentence, while for the system, it's a complete set of micro-structural tasks.

Second, in AIUSD, yield and security are guaranteed simultaneously: The AIUSD itself is pegged 1:1 to USDT, while the yield is carried at the sAIUSD layer, sourced from delta-neutral funding fee strategies. This set of strategies has been running in real trading for two and a half years without any monthly-level drawdowns. This structure allows stability and yield to no longer be mutually exclusive.

Another very important point is that AIUSD offers a truly unified wallet experience: Users do not need to care about which chain they are on, do not need to switch wallets, do not need to deal with bridges, and do not need to learn cross-chain syntax. Cross-chain, spot, perpetual, and payments are all completed within a unified account semantics. This is what we have always emphasized as a Machine-Native underlying structure: money is unified, execution is unified, and the experience is naturally unified.

In this model, our users are not passively "choosing strategies"; what they possess is more like a "money" that can think, execute, and manage, backed by a whole AI financial team they carry with them: an Analyst responsible for deep research, an Execution Trader responsible for order placement and micro-structural execution, and a Wealth Manager responsible for position management and allocation.

We call this experience Vibe Trader or Vibe Coding. Simply put, users express their intentions through semantics and feelings, and AI realizes the complete closed loop of "intention → strategy → execution," making the entire process a very natural financial interaction.

Deep Tide TechFlow: "Stablecoins usable by AI" is an important concept of the AIUSD product. Why is it crucial to achieve AI-driven stablecoin trading? What innovations will this unleash for on-chain finance?

Yao:

Because the executors of next-generation financial activities may not necessarily be humans.

AI Agents will hold assets, make trades, settle yields, and reinvest. However, today's stablecoin systems are designed for humans, not for intelligent agents. What AIUSD aims to do is upgrade stablecoins to "AI-native Money."

Bill Sun:

We believe this will bring several new possibilities:

First, AI can actively manage capital flows. For example, it can automatically switch from AIUSD to sAIUSD based on market fluctuations and then back to liquidity accounts.

Second, autonomous financial networks can form between intelligent agents. For instance, one AI can hire another AI for data analysis and pay for it, with the entire closed loop occurring on-chain.

Third, it will transition finance from "human placing orders" to "intention execution." Users only need to express their goals, and AI completes pathfinding, execution, and settlement through AIUSD.

In the long run, AIUSD is not just a stablecoin but the "settlement hub" of the entire AI financial ecosystem.

Deep Tide TechFlow: Recently, the trading competition among the world's six top AI models has sparked widespread discussion. How do you view this trading practice? If AIUSD were to participate, would it perform better?

Yao:

I see the focus of this competition a bit differently. I think whether the models perform well is secondary; the key is that everyone has limited AI capabilities to "generate judgments" without providing them with a truly complete execution environment. The AIUSD system, on the other hand, has been designed around real trading from day one.

The AIUSD Agent does not just provide opinions; it is directly connected to real liquidity, funding fees, cross-venue execution paths, and clearing systems. Our internal execution and routing engine has been running in real trading for over two years, with an annual trading volume in the trillion-dollar range, and all yields come from structural funding efficiency differences rather than betting on direction.

So if this competition allows for real execution in the future, I would be very willing to participate because that would truly showcase the "AI financial system" rather than just "AI models."

Bill Sun:

I believe this competition indeed demonstrates a fact: models can now understand market signals and form their own trading logic, which is the most significant progress AI has made in the financial sector over the past few years.

However, my first reaction was that this type of competition is fundamentally separated from real trading. The AIs in this competition are all "able to think but not do": they cannot cross-chain move funds, cannot place real orders, and cannot interface with clearing systems, so they can only remain at the level of paper trading.

In the real market, the challenge has never been predicting market trends but rather executing, risk control, funding paths, and clearing—these trading infrastructures. If future competitions evolve into "AI + real execution" contests, I believe the significance would be entirely different. That would test whether a system can survive in the real market, rather than just seeing which of dozens of models outputs the most appealing probabilities.

Deep Tide TechFlow: As of now, what key data achievements has AIUSD accomplished?

Yao:

We have been running the execution and settlement chain for over two years: The underlying strategy stack has achieved an annual trading volume in the trillion-dollar range. This is not a beautification of cumulative matching volume but a real trading execution scale that is reusable, back-testable, and auditable. The core funding fee and micro-structural strategy combination has a historical Sharpe ratio of about 22, with no monthly drawdowns in the past 2.5 years. Additionally, in terms of stability, AIUSD maintains a 1:1 USDT backing, and the yield of sAIUSD occurs at the staking layer, with the two ledgers separated, ensuring that redemption and interest calculations do not interfere with each other, while the custody side is deeply integrated with CEFFU.

The product experience is also becoming smoother: The natural language intention breakdown can now cover cross-chain transfers, micropayments, and combinations of spot/perpetual instructions, allowing users to manage collateral within one account without having to move bricks back and forth across different venues. For me, the real achievement is not a pretty number but that this system has never dropped the ball even in the most challenging market moments.

Bill Sun:

What we are doing now actually evolved from Alpha.dev. That was an AI platform for cryptocurrency news, sentiment, and trading signals, which now has about 3.5 million users and has surpassed 60 million interactions. This data indicates a simple yet real demand: people genuinely want AI to help them read information flows, filter noise, provide insights, and even directly find trading opportunities.

Our quantitative engine is another line, serving as the underlying execution system, capable of achieving a trading volume equivalent to one trillion RMB throughout the year. Our core strategy is a completely delta-neutral, fully hedged funding fee arbitrage, meaning we do not bet on direction but rely entirely on structural funding efficiency differences to make money. Over the past three years, this strategy has averaged an annual return of over 20%, with a Sharpe ratio around 22, and has not experienced any monthly drawdowns. Regardless of how dramatically the market changes, our yield curve remains very stable.

Unlike many "rebalancing" approaches, we focus on dynamic funding fee optimization, tracking, switching, and allocating funds among the top twenty cryptocurrencies by market cap. It is not a strategy that merely appears stable on the surface but one that has truly withstood the test of high volatility environments.

Another noteworthy aspect is our AI analysis and execution framework, which can move from "proposing research hypotheses" to "automatically searching for data validation," then to "generating strategies and executing," and finally even automatically backtesting, creating a closed-loop process. As mentioned earlier, I have applied this architecture to the stock market, using an "AI does deep analysis, I make the PM final judgment" approach, and over the past year, my personal account has achieved a return of about 880%, effectively validating this method in another asset class.

From 1 to 10: Becoming the Universal Settlement Layer of the AI Economy

Deep Tide TechFlow: With the core vision of "AI-Driven Financial Intelligence Future," from 0 to 1 and then to 100, at which stage do you think AIUSD currently stands in achieving this vision? What phase challenges will it face in the future?

Yao:

I believe we have successfully passed the stage from 0 to 1: 0 is proving that this is technically feasible, and 1 is turning it into an account layer that anyone can use—just speak, and the system routes and settles on its own.

And now, I believe we are at the turning point from 1 to 10, where we will focus on delivering the same infrastructure stably to a broader C-end and developer ecosystem.

Bill Sun:

Regarding future challenges, I categorize them into three types:

The first type is the uncertainty of scaling, including funding fee congestion and capacity limits. The larger we grow, the more restraint we need; we would rather sacrifice some yield to keep drawdowns pinned to the floor.

The second type is the compliance puzzle across multiple jurisdictions. Stablecoins, payments, brokerage, and yields—each module's boundaries need to be clearly defined, which requires time and patience.

The third type is the usability of the Agent ecosystem. Intent expression, permission granularity, rollback, and audit logs must be made readily usable for developers. I emphasize refining these invisible technical details to the point of being seamless.

For example, we can liken what we are doing to Stripe: we are building bridges for the digital financial world, converging fragmented rails into elegant APIs, and AIUSD aims to become the universal settlement layer of the AI economy.

Deep Tide TechFlow: Stablecoins and RWA are the absolute protagonists of this cycle. Are there more expansion plans for AIUSD regarding stablecoins and RWA in the future? What changes do you think the continued development of stablecoins and RWA will bring to the global financial market?

Yao:

First, let’s talk about the directional planning.

The positioning of AIUSD has always been "universal stablecoin infrastructure," not "a passive mapping of a certain basket of assets." Therefore, on the stablecoin side, we will maintain a two-layer structure: AIUSD itself will keep a 1:1 backing with USDT, with a simple and clear redemption path; the yield side will be placed on the optional sAIUSD, which will carry "α" through delta-neutral funding fee strategies, without exposing the principal to external credit and duration mismatches. We will continue to deepen this line, including more granular risk limits and dynamic downgrading mechanisms.

Regarding RWA, we will proceed cautiously. I do not oppose bringing real-world cash flows on-chain, but we must honestly face three issues: liquidity layering, verifiability of valuation/pricing, and the legal boundaries of "beneficial rights/ownership." As long as these three issues can be resolved in certain categories, such as the shortest duration receivables or highly liquid government bond opportunities, we will pilot in a retrievable, low-correlation manner, ensuring that RWA does not contaminate the redemption certainty of the AIUSD body.

From a macro perspective, the continued development of stablecoins and RWA will raise the "settlement speed" and "proof of ownership" of financial markets to a new baseline. Once the programmability of global capital is enhanced, pricing will be faster, mismatches will be more expensive, good risks will be priced more accurately, and bad risks will be harder to hide off-balance sheet. I believe this will force the financial industry to return to "tangible assets and compliance capabilities."

Bill Sun:

Stablecoins correspond to the M0 layer, which is the global cash layer. They address the most fundamental question: can the US dollar truly achieve global reach and settle on-chain at any time without relying on local banking systems?

RWA is one layer above this, which is the asset layer, allowing high-quality dollar assets, government bonds, notes, and even private credit to become 7×24, sliceable, and globally tradable assets.

Without RWA, the implementation of stablecoins is likely to occur only in high-inflation, relatively open capital economies. However, once RWA opens up, institutional-level assets can truly face global retail, which is a completely different significance.

The positioning of AIUSD is not to create a new on-chain stablecoin but to make the interaction between different stablecoins a platform layer, allowing users to buy any on-chain RWA with the same experience, whether on Ethereum, Solana, Base, Sui, or Tron. The underlying VM and gas differences are automatically abstracted by the system.

From a macro perspective, this will redefine the baseline of the entire financial system: settlement speed will be upgraded to real-time, proof of ownership will become a default configuration; good risks can be priced more accurately, and bad risks will be harder to hide off-balance sheet; the entire system will concentrate towards "high-quality assets + high transparency + high liquidity."

If I were to summarize this in one sentence: Stablecoins are the global projection of the US dollar, while RWA is the global projection of high-quality assets outside the US dollar. What AIUSD aims to do is to serve as the intelligent connection engine between these two layers, making the flow of money and assets simple, intelligent, and automated.

Our ultimate vision is a very natural thing: a money based on stablecoins + a global exchange based on RWA. You only need to speak to the money, and AIUSD can help you buy any tokenized asset in the world. These assets can also be programmed, allowing for semantic interaction by humans and autonomous invocation, reallocation, and settlement by AI.

Deep Tide TechFlow: In building an AI-driven financial intelligent future, are there more product matrix plans beyond AIUSD?

Yao:

Yes, but we do not plan to "expand the table" by stacking product lines. We are more like building a highway for stablecoins and then filling in the key service stations one by one.

In the short term, we will enhance the "account abstraction + collateral credit" capability, allowing the same collateral to operate efficiently across multiple venues while writing the settlement paths and extreme states into the contracts.

In the medium term, we will consider productizing the yield curve, for example, providing different duration and volatility tolerance tiers above sAIUSD, allowing users and AI agents to select tiers based on "target drawdown/target volatility," rather than passively accepting an averaged yield line.

In the long term, I hope what we create is not multiple apps but a stable, transparent, and accountable financial operating system.

Bill Sun:

Yes, we will not horizontally stack more product lines; all capabilities will converge towards AIUSD as the core. Initially, it will be a hub for stablecoins, and then it will evolve into an intelligent brokerage: regardless of what stablecoin users hold, they can be used, traded, and allocated to different assets within the same experience.

We will also integrate Alpha.dev's AI news, event recognition, and research capabilities, allowing the system not just to read information but to understand your preferences and risk habits, then proactively recommend opportunities that you might be interested in and can trade directly.

Another line is connecting TradFi and CEX. Many people's money is still in Interactive Brokers, Robinhood, and various centralized exchanges. If you want to engage in real trading, you cannot avoid these systems. Therefore, we will directly integrate their APIs into AIUSD, unifying identity, keys, permissions, and order placement into one account layer, allowing users not to split assets or switch systems.

For developers, we will standardize the "intention layer + permissions/limits/whitelists/rollback/audit" into a standardized SDK, while also providing B2B routing and settlement white labels. In simple terms, this means transforming the entire financial interaction from multiple entry points, accounts, and logics into a unified operating system.

Post-AGI Era: Humans as Narrative Producers, AI as Execution Machines

Deep Tide TechFlow: As AIUSD continues to realize the vision of "AI-Driven Financial Intelligent Future," could you share your perspective on what trading in the Post-AGI era looks like?

Yao:

In my view, trading in the Post-AGI era will not feature fancier interfaces but rather quieter infrastructures. Price discovery will be faster, and prediction itself will no longer be the primary source of returns; true excess returns will come from orchestrating capital, computing power, data, and clearing paths with minimal friction. Trading will shift from "humans monitoring the market" to "intention governance": you declare goals, constraints, and credible boundaries, and the system executes, reviews, and corrects within a verifiable track. The human role will resemble that of a governor and narrative producer; we set the rules and boundaries, define what constitutes acceptable risk and desirable returns for the system, and then delegate most of the execution to machines. Capital will move faster but also more methodically, with each movement carrying auditable reasons and accountable signatures. This is the trading world I envision: efficiency is maximized, but restraint and explainability are also embedded in the protocols.

Bill Sun:

My definition of AGI is quite pragmatic; simply put, it can systematically replace white-collar jobs, allowing various professional capabilities to gradually converge within the same model, rather than the sci-fi notion of a "sudden singularity." Given the current pace of progress, I believe the so-called Post-AGI phase may be observable within three to four years.

At that stage, a particularly critical aspect will be that AI needs a set of "money" that it can truly understand and directly use. This is the direction we set for AIUSD—making it the underlying currency for M2M, or Machine-to-Machine payments. This way, AI can incorporate "money" into its world model: paying for computing power, data, services, and also investing and reinvesting, treating finance as part of its actions.

At the same time, I believe two very important forms of intelligent agents will emerge: one is the AI Scientist, which uses AIUSD to buy computing power, data, RL environments, and energy, forming a true self-improving loop that can iterate on itself; the other is a free market among multiple intelligent agents. We will not enter a "planned economy of a single super model," but rather a scenario where countless independent AGIs trade, compete, and collaborate using AIUSD, completing resource allocation through price discovery.

In such a world, the significance of AIUSD for humans and for AI will be different but complementary. For humans, it is a Financial OS that allows you to manage wealth through intention governance, without needing to understand all the details; you only need to express your goals, and the system will handle strategy, execution, and settlement for you. For AI, it is an Agentic Money that truly empowers intelligent agents with financial agency, allowing them to operate in the economic system as participants rather than bystanders.

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。