Original source: OKX
OKX Ventures recently held an online sharing session (Twitter Space) themed "Autonomous Agents Reshaping DeFi," delving into one of the most exciting cross-sections in Web3: the rise of DeFi autonomous agents.
This discussion went beyond the early hype surrounding AI chatbots and confronted the core issues: How do autonomous agents create real value, manage risks, and fundamentally reshape the user experience in decentralized finance? To gain frontline insights from builders, we invited four industry pioneers who are dedicated to shaping the future of Agentic Finance:
• Sam, CEO & Founder of Cambrian Network
• Neo, CEO & Founder of Almanak
• Renç, CEO & Founder of Giza
• Colin, Product Lead of Makina
AMA Summary:
AI is not a gradual improvement of DeFi, but a paradigm shift. The goal is to transform DeFi from its current complex, product-centric model into a simple, user-centric personalized service. Ultimately, this aims to enable users to achieve their financial goals autonomously, without deep technical knowledge.
Clear division of labor for AI: "off-chain brain," not "on-chain hand." The current role of AI in DeFi is strictly defined. It primarily acts as an "off-chain brain" for complex reasoning, data analysis, interpreting user intent, and generating deterministic and verifiable strategy code. AI itself does not directly touch or manage on-chain funds; the final execution is based on auditable, traditional finance-like logic.
Safety first: Managing risks through "human oversight + technical guardrails." We need to prioritize safety and risk control to address user concerns about AI going out of control. The core solution is that AI operations must occur within "guardrails" preset by human risk managers and enforced by code, and the generated strategy code must be fully auditable and verifiable by humans. This ensures that AI's decisions are controllable and traceable.
Serving two types of clients: enhancing efficiency for institutions and lowering barriers for retail users. The product targets both institutional and retail users but in different ways. Institutional clients like hedge funds and DAOs significantly reduce the cost and time of strategy development and operations through AI. For retail users, the pursuit is "complete abstraction"—hiding all the complexities of DeFi, allowing users to simply express basic financial goals (like "I want to earn stable returns"), leaving the rest to the agents.
Ecological synergy: The application layer and infrastructure layer develop together. The realization of Agentic DeFi requires a complete ecosystem. This includes not only user-facing strategy applications like Giza and Almanak but also "rails/settlement" layers like Makina that provide secure, cross-chain execution environments, and infrastructure layers like Cambrian Network that supply reliable, verifiable data "fuel" for agents.
Ultimate goal: Democratizing professional financial strategies. Through AI agents, the aim is to break down the barriers that limit access to complex quantitative strategies in traditional finance to a select few. Strategies that would typically require millions of dollars and months of development time for hedge funds will be made available to everyone at a very low cost and with great speed, truly achieving inclusive finance.
AMA Questions and Discussion Transcript:
1. Introduction to the product and main focus
• Sam (Cambrian Network):
My career began at one of the national laboratories in the U.S., working in cryptography primarily focused on reverse engineering cryptographic hardware. I then obtained a PhD in reinforcement learning from the University of California, Santa Barbara. After that, I founded my first company, Semiotic Labs, where we were the core development team for The Graph protocol, focusing on AI, verifiability, and The Graph's payment system.
During that time, we did a lot of work related to agents. For example, we released the first reinforcement learning agents for dynamic pricing within The Graph in 2022. In 2023, we launched the first publicly available blockchain data terminal, allowing users to query real-time and historical data using natural language to generate SQL. By 2024, based on these experiences and our firm belief that AI will have an immediate and significant impact, and that cryptocurrency will become increasingly important in the global economy, we decided to incubate Cambrian from Semiotic. Cambrian focuses on providing on-chain and off-chain financial intelligence. Providing this intelligence to agents is our beachhead market.
• Neo (Almanak):
I have been in this field for nine years. Before founding Almanak, I ran an agency providing data science and consulting services for DeFi, trading, and crypto asset management, so I am very familiar with how this field operates.
Regarding Almanak, we have been in the market for four years. We like to call ourselves a vibe coding company; you can think of us as the Cursor of the DeFi space. Essentially, we use AI agents to discover and build complex trading and asset management strategies. These strategies are fully verifiable deterministic code. You can think of these strategies as the same as those used by any hedge fund for trading.
• Renç (Giza):
I have a background in product and marketing, and before founding Giza, I spent five years as a product lead at Johnson & Johnson. During that time, I also built smart contract systems across various financial use cases. I am fortunate to have a team with backgrounds in machine learning and data science, so we approach this from the perspective of machine learning and AI with financial experience.
Since 2022, we have been building Giza. Giza builds agent applications for automated finance—these autonomous systems can execute complex financial strategies on behalf of users and institutions, with zero operational overhead. I would say this is our version of "banking the unbanked." In our view, financial exclusion is not just about whether you have a secure account to store your constantly inflating fiat currency, but also about being cut off from various opportunities. Can you respond flexibly to changing markets, seize these huge opportunities, and avoid risks when necessary? This is the question we want to answer. Our work at Giza is about democratizing all these capabilities.
• Colin (Makina):
I am responsible for the product at Makina. I joined the team about four months ago. I have been in the crypto space for over ten years. My background is initially in traditional finance, and I started getting involved in DeFi around 2016, building products since then.
At Makina, we focus on institutionalizing what we call "DeFi execution." Beyond strategies and vaults, we are really interested in creating a secure and reliable way to interact with any DeFi protocol or any EVM. This is crucial for anyone trying to run strategies, whether they are operated by humans in a more traditional way, more passively or automatically, or through AI-driven agents.
We look at this issue from multiple angles. First, we focus on what we call "operators." This is similar to the "curators" you see in other protocols. They can trade securely while having control over what they can and cannot do. In addition, we also use AI extensively to improve user experience, such as providing better recommendations, better understanding what users are doing, and exploring different ways to integrate new protocols to ensure that whether humans, agents, or other types of algorithms are operating the vaults, they can quickly get up to speed and create maximum value in a secure manner.
2. What inspired you to start your current project? Why do you believe AI will bring value to your product, and what is the main value proposition?
• Sam (Cambrian Network):
I completed my PhD in December 2019. Reinforcement learning is very popular now, but in 2019 we were in a bear market for reinforcement learning. This was one of the reasons we initially focused on fully homomorphic encryption when we started the company.
But when GPT was released in 2022, I was initially shocked like everyone else. However, I actually thought we were at the beginning of a bubble—I know many people today also think we are in a bubble. But by 2023, a year after GPT's release, I kept seeing progress, and I developed a profound belief, which I still hold today, that we are at the beginning of a new revolution, the last one being the internet revolution. Before that, we had the Silicon Valley revolution, and going further back, the industrial revolution, and so on.
So, we are in the early years of a new revolution that will not disappear. I encourage everyone present to be prepared; AI's capabilities will double every year for the foreseeable future, impacting everything and every aspect of our lives—it has already begun.
In addition to this belief, I also got involved in DeFi. As early as 2021, my previous company created Odos.xyz, which we spun off. It is a DEX aggregator. So I have a deep belief in the financial freedom and literacy that financial applications and cryptocurrencies can bring.
During the pilot projects and experiments mentioned in our introduction, one of the most difficult things I noticed was that data and information about what is happening on-chain, as well as other relevant information crucial for on-chain and off-chain financial decisions, were very difficult to obtain. And this is critical for financial decision-making. This is why we focus on Cambrian. We believe that every project engaged in agentic finance or autonomous finance needs reliable, fast, comprehensive, and verifiable information to feed their agents. This is crucial for the success of these projects, which is why we decided to focus on financial intelligence.
• Neo (Almanak):
We like to call ourselves "AI for DeFi." Regarding inspiration, Almanak was initially a company that used AI to optimize trading and asset management strategies. We have always worked with large asset management firms and major allocators, so we always have access to large capital. Almanak has been established for four years.
Three years ago, when the ChatGPT craze began, we knew it would become very important. So we asked those big clients, "Hey, what would make you confidently entrust your money to AI management?" They said—these are particularly large capital allocators—that they would never deposit more than, say, $100. They are very afraid of AI manipulation, indirect prompt injection, and various "unknown unknowns." Or simply put, when money is lost, they want a person to sue.
So, in conversations with these institutions managing billions of dollars, we asked ourselves, "Well, what is AI best at?" Today's AI excels at coding. Its coding speed is hundreds of times faster than that of an average person. AI is also very good at reasoning; it processes information a trillion times faster than humans.
Thus, we seized these two characteristics and applied them to Almanak. We created an "Agentic Swarm"—or a team of agentic swarms—whose objective function is to write high-performance strategies, discover market opportunities, handle market dynamics, optimize existing strategies, and feed all this information back to users.
In our ecosystem, AI collaborates with users. It provides you with ideas about strategies, ideas about optimization, and ultimately realizes the code. However, if something goes wrong, the hedge funds will look for you. What we have created is a method that can reduce the time to develop any complex financial strategy from months to minutes. Moreover, we have reduced the cost of developing such strategies from millions of dollars to just a few dollars, or even less than ten dollars, depending on the complexity of the strategy.
Once this strategy is created, it is exactly the same as any strategy used by hedge funds. It is deterministic, verifiable, and you can backtest it, simulate it, and deploy it—so you know what will happen. AI will never touch your funds. AI merely enhances the strategy creation and discovery process but never touches the funds. So far, this approach has been effective. We have seen the confidence of those large allocators, and our current total value locked (TVL) is $160 million.
Equally important, once a deterministic Python strategy is created, you can wrap it into a vault. These vaults are fully composable—you can place them on Pendle, Curve, and so on. So this is also very cool. We like to think we have created a new asset class called "tokenized AI vaults." Again, AI never touches the funds, so large allocators feel very secure depositing here. They know who to look for—they will look for you, the vault operator—and you simply use Almanak as a tool that is 100 times faster at coding and a billion times faster at cognition.
Additionally, as Sam mentioned, we also focus on creating financial agents. Our agents are fine-tuned for quantitative reasoning capabilities, so we ensure these agents are as smart as, or even smarter than, any other quantitative analyst in the industry. But our inspiration mainly comes from close collaboration with large allocators, pragmatically meeting their needs. We simply ask them, "Okay, what do you need? Where will you invest your money?" Then we make it happen.
• Renç (Giza):
As I mentioned, before founding Giza, my partner and I had been building smart contract systems across different financial use cases. One thing was clear: while these self-executing contracts opened up the vision of open finance, we frankly believe that in its current state, the pace of innovation is too slow to remain competitive with traditional finance. This was the main driving force behind our search for ways to transplant complex off-chain computations onto the chain to significantly enhance the capabilities of decentralized systems and the user experience of interacting with the decentralized finance world.
Since 2022, we have been deeply researching verifiable AI. We were doing this before it became trendy, explaining its importance to people, especially in financial use cases. We explored all possible machine learning hotspots and financial use cases in decentralized finance.
For us, the value of AI is twofold. On one hand, general intent processing is key to understanding what users want to achieve financially without technical input and autonomous specialized actions. On the other hand, it concerns executing complex adaptive strategies on-chain in a precise and cost-free manner. This second part leans more towards small, internally developed machine learning models and traditional financial algorithms that are fully explainable, verifiable, and customizable.
• Colin (Makina):
Our story at Makina began with Dialectic. Dialectic is our design partner; we are now independent of them, but they had an insight while establishing their own fund. For those unfamiliar, Dialectic is a very active investor in this field and has been one of the earliest and most advanced participants in the on-chain yield strategy space since 2021. They manage many different affairs through systems built over the years.
One thing they quickly realized is that to compete in this field, to make money, to outperform, and to attract more depositors and limited partners (LPs) into those funds, they needed to outperform other strategies on a risk-adjusted basis. To this end, they built many different tools utilizing scripts. One of the technologies they used is an open-source project called Oiler, to which they also actively contributed. They realized that many of the tools they built would actually thrive better if they became an open infrastructure. This is pretty much the beginning of the Makina story.
We basically pushed it to market, collaborated with them, and are now expanding to other operators in this field. We hope to support future development directions, which are moving towards more automation. This automation will heavily rely on what happens within the blockchain and what happens in the macro environment—wherever you get that information from. And the best way to handle that information is what we have heard from everyone here.
We first tackle the problem from the perspective of DeFi and financial infrastructure issues, then explore where to apply best execution, best decision-making, and best data analysis. And that is clearly autonomous agents.
We realized, as we heard from Neo—by the way, that was a great opening, hearing some of the concerns people have about entrusting money to AI—we handle it in a slightly different way. But we firmly believe that as these technologies improve, people will begin to understand that we can supplement and expand products while addressing some of the major cost issues occurring within the asset management industry. So we are firm believers in DeFi, firm believers in Ethereum, and firm believers in AI and its advancements in these industries.
3. Who are your main clients now? What are their pain points?
• Neo (Almanak):
At Almanak, our product needs to address two aspects of the problem. We must solve the supply issue of complex strategies and vaults. So, we have collaborated with many different DAOs and every curator on Morpho, such as Stake DAO, MEV Capital, Block Analitica, Gauntlet—all of these people. When it comes to DAOs, we are in talks with most of the top 20 DAOs ranked by DeFi Llama. Why would they use our product? They basically create vaults that utilize their assets.
I’ll take the largest asset management participant, Ethena, as an example. Imagine you could have a USDe vault that continuously optimizes and seeks the highest USDe yield across all DeFi protocols. We are in discussions with these people.
We are also talking to many new projects. I don’t know if you’ve been paying attention, but there are many complaints about high FDV token economics right now. So, at Almanak, we also allow projects to use AI to launch their own liquidity provision or trading strategies. Users can simply use our algorithms to start a market or a trading competition.
Last but not least are the ordinary users, as capital comes from them. So I just explained the supply side of the vaults. The supply of capital comes from users. Once these vaults are deployed, anyone can deposit funds and benefit from them, of course, in exchange for sharing some profits with the vault's curators. These vaults will be completely permissionless, so anyone can deploy a vault. But I just want to give you some perspective on who will manage them and who our initial clients are.
Additionally, there are asset management companies and hedge funds. We are in talks with centralized finance (CeFi) entities managing billions of dollars who simply want to automate their deployment systems. Quantitative analysts (Quants) are extremely expensive and hard to find. They can outsource all of this to our agents, rapidly deploy complex trading strategies, and become a hedge fund in a week or even a few days.
I also want to mention an important point here. As a user, you will be able to stake tokens in a contract very similar to ve contracts. So you can vote for your favorite vault, vote for your favorite DAO, or vote for your assets when you deposit into the vault to increase your rewards. Our product is very complex. The supply side of the vaults will be provided by professional users, but the supply of capital is open to everyone.
• Sam (Cambrian Network):
Colin just mentioned the allocation of capital between yield-generating vaults and lending protocols. Our focus is on measuring where these yields are generated. To optimize the types of strategies Colin mentioned, you need to understand the historical yields generated across different chains and the various protocols within those chains. This requires complex data plumbing, tracking on-chain activities, including EVM and non-EVM chains, and monitoring the protocols within those chains.
Builders need both historical information to adjust their strategies and real-time information to execute their strategies. This is one of our areas of expertise—tracking all this information. If you think about an RPC provider, what they provide is real-time raw information, which means the information coming out of the RPC provider is not always clear. What we do is decode all the historical data and, based on our understanding of the protocols, decode the information and start tracking, for example, the yields being generated.
Currently, we are in a closed testing phase and collaborating with the Coinbase developer platform. We are working with Olas to become part of the Olas hedge fund cluster, providing historical and real-time on-chain and off-chain data for agents within Olas.
We are also collaborating with several other projects: we are closely working with Truflation to provide them with sentiment analysis and wallet activity. Another more interesting project we are collaborating with is called AskPire. They are tracking thousands of GitHub repositories related to tokenized projects. We track historical contributions and the quality of contributors, while AskPire is building customized trading strategies using our data, enabling them to correlate project activity with future token prices. So, I hope this gives you a rough idea of the types of information we provide. This is all based on the common needs we see in agentic finance projects.
• Renç (Giza):
To set the stage a bit, at Giza, we are not really interested in incremental improvements. I believe DeFi has long been trapped in one incremental improvement after another. What we want to achieve is a complete paradigm shift in user experience (UX) in the financial sector—not just in Web3, but in the entire financial industry. We want to shift finance from being product-centric to being user-centric. We have a very firm view on this: personalized finance is the way forward.
Our vision is not to create yet another DeFi protocol; it never has been. Instead, we aim to create a companion that operates 24/7, capable of executing and providing insights into your financial situation, helping you achieve your financial goals. This is the North Star we are chasing. Given the robustness of our infrastructure and this tailored personalized finance North Star, Giza's agents are now able to serve both retail users and institutions simultaneously.
The institutions we collaborate with today have more rigorous and complex needs, covering everything from custody requirements to risk frameworks to liquidity directives. Giza was built to meet these needs through customized agent strategies rather than off-the-shelf products. This includes everything from designing custom agents to isolated infrastructure, real-time monitoring, audit trails, and providing white-label implementations for funds, fintech partners, and new banks (which have many proactive demands).
For individual users, I think there are still some areas worth exploring. This might be where we still offer the same complexity but in a less complicated manner. For retail users—i.e., "banking the unbanked"—we can enable them to interact with decentralized finance through an extremely simplified interface that completely abstracts away the strategy layer. We take on the responsibility of financial decision-making for users. We have automated the decision-making process. I believe this is one of the most obvious differentiators for Giza; we have the courage, expertise, and talent to take on this daunting task.
For both retail and institutional segments, we are exploring some unique requirements. In short, retail users want complete abstraction and accessibility, while institutions require higher security, monitoring, and reporting standards. We have the capability to meet both.
Giza has been building a critical asset base, which is the stablecoin market. Clearly, it is not going away anytime soon. Its total market cap has reached $300 billion, and every circulating stablecoin represents potential capital that can be autonomously optimized by Giza agents. This is why we have built our first agent for this space and will continue to expand its coverage and capabilities. Of course, this also allows us to serve treasuries, DAOs, institutional funds—anywhere that can abstract DeFi, anywhere someone asks, "How should I invest in stablecoins?" Giza is there.
• Colin (Makina):
Renç just said some very interesting things about how we can move beyond incremental changes in the financial sector. I think we all got into this technology because we recognize that the traditional financial system is currently not working for people. I believe this is one of the guiding lights for all Ethereum participants.
At Makina, we strive to build security and assurance into everything we do while making it scalable. We firmly believe that by providing this infrastructure, we can deliver the best outcomes for anyone, whether large institutions or small retail users.
Our view of the world is very similar to what Neo mentioned. Some entities have investment needs, and some entities want to meet those needs. We are working hard to ensure that the best managers of financial outcomes have access to tools they can operate safely. We firmly believe this is a growing field.
If we look at the traditional financial markets, there are currently about $150 trillion in assets under management globally. An interesting fact is that about 60-70% of these are actively managed, and that share has been declining. A large part of the reason is that people are paying significant fees without necessarily outperforming ETFs. We have also heard a lot about ETFs in the crypto world. ETFs are also quite revolutionary, changing the views of many traditional finance professionals due to their low costs.
We firmly believe that as we see advancements in technologies like Ethereum, EVM, and AI in terms of security and automation, these costs can be reduced, allowing people to achieve excess returns in a more cost-effective manner through better strategies. This is really important to us on a global scale. It’s not just about doing better for some entity on Wall Street or in the City of London. It’s about ensuring that anyone who needs to achieve financial outcomes can do so.
Moreover, we firmly believe that these should be built directly into DeFi protocols. We should build tools that allow managers to convert these productive assets into collateral or use them in different forms within the DeFi ecosystem. This is the true way to develop the DeFi economy. It can be stablecoins, but it can also go far beyond that, allowing people to match their future liabilities with the assets they have on hand and pass them down through generations. We believe this will fundamentally change the way people achieve prosperity through their own wealth. Again, we are staunch advocates of AI and Ethereum in achieving this goal.
4. In your tech stack, which parts rely more on AI capabilities, and which parts rely less, and why? Additionally, since we are discussing building a financial system that utilizes AI, risk management and control are very important. When you consider AI safety, how do you incorporate risk management or control into your workflows?
• Ray (OKX Ventures):
When people discuss DeFi agents, it seems that many users, especially retail users, still have misunderstandings about the concept. They might think, "Hey, we can directly use AI agents and rely 100% on them to make financial decisions, manage our funds, and find alpha," but that is not the case. In reality, we want to help clients build a financial system that utilizes AI capabilities to some extent, ultimately improving efficiency or decision quality. However, we still need to build a reliable, deterministic, or verifiable workflow because we need a reliable financial system before we invest significant amounts of money. That’s why I want to ask some questions about how you consider potential risk factors in your systems.
• Renç (Giza):
Yes, absolutely. I think this is very critical, as a company that is pioneering the launch of agents that make financial decisions on behalf of users and institutions, one of the biggest challenges we have had to overcome in the past few months is educating the public about the reasonable questions you raised. Will agents run away with my money? Can we explain what AI is doing with our money? To what extent is this 100% deterministic? Will they hallucinate? We have to go through all of this to get people to accept it. Because it is a brand new tool. For Giza, it is not a vault where people are already accustomed to depositing funds; it is something entirely new. Each user has a dedicated agent serving them. In your question, it is important to distinguish or define what "AI" means in this context.
Most of the questions you raised stem from an understanding of LLMs. For us, the ability of LLMs to parse our users' general needs and parameterize them into preferences is quite remarkable—essentially transforming vague, human-level inputs, from "I want to safely earn money on my stablecoins" to "I want to outperform U.S. inflation by 5%" or "I want to take on moderate risk exposure to ETH," into structured financial parameters.
But the part of AI that is typically recognized—that is, LLMs—stops there for us. Once the intent is parameterized, execution shifts to specialized agents built on algorithmic logic and optimization functions, which are deterministic, verifiable, auditable, and capable of continuously self-adjusting across markets and protocols. So, by combining the two, we gain tremendous customizability from the AI side while also obtaining professional, robust, secure, and strategy-constrained execution from specialized agents that excel in financial markets.
• Neo (Almanak):
I’m not sure how AI can be both deterministic and verifiable, but regarding us: how we use AI and how we address security issues.
Again, we are very pragmatic about everything. We don’t want to reinvent the wheel; we just adopt what works and is in demand in the market. We specifically leverage AI to generate code, which is 100 times faster, and this code is deterministic and verifiable. If you ask any hedge fund manager, any quantitative analyst, or any developer if they are familiar with what our agents produce, they will be familiar. If he gets a call from his limited partners (LPs) or a bank saying, "Hey, can you show me the code?" he can showcase the code. If money is lost, you will be able to say who stole the money and how it was stolen because the code has vulnerabilities or other issues. So this is very important. Security is as safe as any other hedge fund or bank.
When it comes to execution and conception, we use agents in a way that is very similar to others here: basically screening the market, finding alpha, identifying the best solutions, finding the best trades, simulating strategies, backtesting, and simulating trades to avoid slippage. So AI collaborates with you in conception, but ultimately, you make the decisions. You decide whether to implement the strategy provided by AI; you decide whether to update the code. The code is completely verifiable and deterministic. Again, we just adopted effective methods and made them 100 times faster in coding and a billion times faster in reasoning.
When it comes to the blockchain infrastructure layer, we also don’t want to reinvent it; we just adopt what works. We want everything to be fully composable, so we use composable vaults. Security is addressed through transparent permissions. Each vault has transparent permissions, so you can see on-chain what this vault can access. Whenever someone—be it the vault manager or curator—changes these permissions, it will be visible and transparent. This completely replicates the practices of hedge funds.
Additionally, we have created—what I think is very unobtrusive but one of the most technically sophisticated and challenging things we have done—structured workflows for our agents. We currently have 18 agents; now everyone can use 7 of them. These agents operate like quantitative analysts, but they function on infrastructure similar to traditional hedge funds. We have drawn from what traditional hedge funds possess—the infrastructure for creating, backtesting, simulating, and optimizing strategies—but we are not creating it for humans; we are creating it for AI. So even the creation process itself is as rigorous as any other hedge fund.
Basically, we only use AI in areas where it is not critical to the loss of funds. Because of this, people feel secure depositing funds, and we have received numerous proactive requests from funds and asset management companies to use the tool. I would say our security is as safe as the blockchain.
• Colin (Makina):
I have heard a lot of very interesting things from these friends, and I may not be able to debate the determinism of AI, so I will leave that to you.
Once again, we are approaching this issue from a financial perspective. In answering the first question about where we use AI, I want to emphasize that this is how we are using it today. I am not an expert in the inner workings of these AI agents; we strive to provide tools for those experts. We have seen an evolutionary process. Anyone here who has used AI in any capacity has seen tremendous progress in a short amount of time, and this progress will continue.
The areas where we currently rely on AI are in automation. Of course, as we heard from Renç and Neo, you need to set up guardrails for this. One very interesting aspect of Makina is that we have brought these guardrails to cross-chain. L2 is an important component of Ethereum, and the alternative L1 of EVM is as well; we maintain the same controls when transferring assets across chains. This means we can open up new investment areas and absorb a large amount of information. Sam mentioned some great different sources of information provided by his company. Being able to read what is happening on social media—I mean, we are all on X (Twitter) now—is crucial.
Most of us may have intentionally or unintentionally spent some time this week thinking about Monad. A lot is happening on Monad, and getting in early will help some people outperform others. But you shouldn't do this without control. This is what we are really introducing. We believe AI will play an important role in deciding when and where to deploy funds, but not without control.
• We firmly believe that at this point in time, these controls still need checks and balances. We have a "Risk Manager" role within the vault. What this really means is that someone can decide based on a whitelist what an operator—be it an agent or another person—can access. This is encrypted and stored in every blockchain activated within our machines or vaults. So when an operator makes decisions, timing, direction, magnitude—all of these can be determined by these operators in various ways. But limiting access is something that requires more consideration, and we need to maintain this control. We specifically allow the Risk Manager to use AI tools for rapid iteration and to build what we call blueprints or scripts, but ultimately, for now, the final decision is still made by humans.
On the other hand, from the user's perspective, we are experimenting extensively to better understand what kind of recommendations our deposit users want. This moves away from the question of what decisions to execute and more towards understanding what users are trying to achieve and helping them match with what is available. As I mentioned, I have worked in traditional finance, and anyone who has spent time in traditional finance truly understands how difficult it is to obtain information. We want to help people get the information they want and better understand performance based on their own intuition and goals. We believe AI is a very good tool to achieve this. It is not perfect. We have been experimenting with our FAQs, which are run by an LLM bot. The team will tell you it still needs more tuning and more data input. But our users appreciate this very much, and it helps us adjust our own user interface and front-end experience to better serve these users, allowing them to understand exactly what they want to know in a very efficient way without having to read 18 pages of FAQs.
Another thing I really want to emphasize is that we do not use AI to write smart contracts. We have top-notch Solidity developers. We are working extensively with auditors to ensure that everything underlying is secure. We believe that, at present, all of this should be done by very experienced humans, and we are very satisfied to have experienced individuals on the team.
• Sam (Cambrian Network):
At a high level, I want to share how I view agents. I think intelligence exists on a spectrum. The agents you see deployed today, I categorize them as algorithmic agents. The decision-making strategies of these agents are deterministic; they are mathematical, use optimization, and operate entirely according to the creator's intent.
On the other end of the spectrum, we have AI agents. The most advanced AI we have today are LLMs. LLMs are creative and can adapt to different conditions. However, the problem with AI agents today is that they are non-deterministic. You can give GPT the same prompt, and each time you run it, you will get different answers. Besides non-determinism, they also frequently make mistakes; they can hallucinate.
The promise of AI agents lies in their adaptability, far surpassing algorithmic agents. I believe we will see—and I am very confident that—the determinism issues of LLMs will be resolved. For example, there is a company called Sakana AI that was incubated from Google Brain; they recently published results showing significant progress in getting LLMs to generate the same content every time. I believe EigenLayer will also release similar work. In terms of improving accuracy and hallucinations, you can assume that the error rate will be halved every year on any significant task.
So to summarize, now, as Renç said, LLMs are very good at capturing intent and transforming that intent into parameters that can be input into algorithmic agents, which can then operate reliably. On the other end of the AI spectrum, you can assume their performance will double every year, and they will become active decision-makers in managing our financial decisions.
Now, specifically regarding what Cambrian is doing, I am very concerned about data. In terms of data issues, we ensure accuracy by using cryptography to verify that all our inputs are correct. If you start trying to obtain blockchain data, you will find it is often incorrect. Cryptography is the solution to ensure its correctness. This raw data then enters a database, and we begin tracking things like yield; we must ensure that our yield tracking algorithms align with how the smart contracts of all the protocols we track are written. So we have to conduct extensive cross-checking and testing with other sources.
OKX Ventures Thesis on DeFi Autonomous Agents
• Ray (OKX Ventures):
According to our previous research, the DeFi Agent track is undergoing a critical turning point from conceptual frenzy to reality testing in the second half of 2024. The first wave, dominated by the "GPT Wrappers/Chatbots" model, promised users the ability to easily navigate complex DeFi operations through natural language. However, this seemingly beautiful vision quickly exposed its fundamental flaws in practice.
These early "DeFAI terminals" generally encountered three major dilemmas in practical applications: first, LLMs struggle to accurately identify highly complex and personalized user intents in financial scenarios; second, the industry lacks supporting tools to reliably convert vague intents into precise on-chain operations; and finally, even when users have powerful tools, they often find themselves in a state of "decision paralysis," not knowing what instructions to give.
However, the deep-rooted commonality of these problems lies in the fact that the first-generation agents attempted to rely entirely on non-deterministic LLMs to dominate the entire process from intent understanding to transaction execution.
This fundamental paradigm flaw led to a rapid reshuffling of the market. Faced with extremely low actual conversion rates and poor user experiences, the vast majority of projects have thus disappeared. The survivors have shown clear route differentiation:
• Some projects attempt to make incremental improvements at the UI level, optimizing prompt engineering, but this does not address the core issues.
• Meanwhile, another group of projects, which are truly leading the market direction, has chosen a more thorough transformation—they no longer insist that AI directly understand everything but instead focus on specific scenarios, providing clear value to users through pre-built workflows.
These emerging autonomous agents build real depth capabilities through preset, validated processes and focus on the DeFi Adapter Layer and Cognitive Engine, thus clearly shifting the market focus to the latter, ushering in the era of autonomous agents. To understand the essence of this shift, we must first clarify the fundamental differences between the two execution paradigms.
We believe that a secure, reliable, and scalable AI financial solution must abandon the model of having LLMs execute directly and shift to a structured workflow centered on "determinism"—for any given input, the system always produces the exact same output. This is akin to a mathematical formula or a piece of traditional computer code, whose behavior is predictable, verifiable, and reproducible. This workflow should adhere to the following four core principles:
Curated Data Sourcing & Environmental Isolation: The channels through which agents obtain external information (such as market data, on-chain data) must be strictly vetted and formatted API connectors, rather than allowing them to scrape freely from the open internet. This fundamentally eliminates security risks caused by data pollution.
Pre-vetted Strategies, Not Ad-Hoc Decisions: Any trading logic cannot be an AI's improvisation. Each strategy must be developed, rigorously backtested, and simulated in a sandbox environment before deployment, with its goals and behavioral boundaries "solidified" before entering live trading, ensuring its behavior aligns with expectations.
Permissioned Execution & Risk Boundaries: Strategy execution permissions should be strictly limited. Clear boundaries of rights and responsibilities should be set through smart contracts (e.g., only able to interact with whitelisted protocols, strict fund transfer limits, etc.), ensuring that even in the worst-case scenario, potential losses are contained within a controllable range.
Continuous Monitoring & Circuit Breakers: Once a strategy is live, it must be monitored in real-time by an autonomous risk management system operating around the clock. If the strategy's behavior deviates from expectations or the market experiences extreme volatility, this system should be able to immediately initiate a "circuit breaker" mechanism, taking intervention measures such as reducing positions or pausing the strategy, acting as the ultimate "safety valve."
It is precisely due to the paradigm flaws of the first-generation products that the market underwent a rapid reshuffling. Faced with poor user experiences and extremely low conversion rates, the vast majority of projects have disappeared. The survivors have shown clear route differentiation: some projects remain at the level of incremental improvements in UI, while those truly leading the direction are the "Autonomous Agents" that have chosen a complete transformation. Do not misunderstand this concept; these emerging agentic products no longer insist that AI understand and do everything. Instead, they provide clear value to users in specific scenarios through pre-built, validated workflows. They focus their R&D efforts on building a truly defensible DeFi Adapter Layer and Cognitive Engine. As a result, the market focus has clearly shifted to the latter, ushering in the era of Autonomous Agents.
Conclusion:
Although the Crypto x AI track is heavily questioned, we firmly believe that, under the premise of adhering to the principles mentioned above and appropriately leveraging LLM capabilities, this field can bring highly attractive value propositions, especially for institutional clients. This includes enhancing multidimensional information analysis capabilities (capturing complex factor correlations that traditional algorithms struggle to cover), significantly improving code development and deployment efficiency, and achieving more powerful automation execution capabilities. Therefore, we are willing to pay long-term and continuous attention to the development of this field and seek early teams that align with our core principles.
This article is from a submission and does not represent the views of BlockBeats.
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。
