Original title: DeFAI: The Rise of Trading Agents
Original author: Defi 0x Jeff, head of steak studio
Original translation: zhouzhou, BlockBeats
Editors note: This article discusses the development of DeFAI and introduces how platforms such as Cod 3 x and Almanak allow users to create, customize and run autonomous trading agents to improve the efficiency and privacy protection of trading strategies. The article mentions that these agents are developing rapidly, but also points out that there are still many immature projects in the market. In the coming months, as the actual performance of Cod 3 x and Almanak is revealed, the competition for trading agents will gradually intensify. The author also expressed his expectation for testing different trading strategy platforms and plans to share relevant updates.
The following is the original content (for easier reading and understanding, the original content has been reorganized):
An agent makes money for users autonomously, researches and identifies high-quality options, analyzes sentiment and on-chain data, and executes trades based on its signals and insights. No more spending hours looking for excess returns like a human, avoiding phishing scams, and avoiding rug events. You can have an agent that continuously generates excess returns for you at any time. This is the vision of autonomous trading agents.
Origins: The InvestmentDAO narrative
Autonomous trading agents have experienced rapid growth, initially stemming from the InvestmentDAO narrative — fund managers raised funds through daosdotfun on Solana with the goal of generating excess returns by investing in meme and AI agent tokens.
ai16z dao was the first DAO to start this narrative (they later shifted their focus to focus more on the open source innovation movement with ElizaOS).
Since then, we have seen the emergence of many DAOs, in particular:
Big Pharmai: Invests in DeScI and has its own BADDIE framework
·AimonicaBrands：Waifu VC, aiming to go beyond traditional venture capital.
sekoia virtuals: Autonomous venture capital using AI agents for investment analysis, focusing on Virtual agents.
Vader AI: Invested in DAO/DAO creation platform, invested in Virtualsagent.
Wai combinator: An investor and incubator that actively invests in micro virtual agents.
aicceleratedao: Investing in and developing DAOs to accelerate the progress of DeAI across different blockchains.
All of these are thriving on platforms like daosdotfun and daosdotworld. While the InvestmentDAOs ecosystem has grown into a vibrant network that supports agents for marketing/exposure and maintains the price of agent tokens through the Agent (3, 3) strategy (DAOs buying other agents or DAOs supporting each other), investors trust InvestmentDAOs to outperform the market and earn excess returns.
Coupled with the fact that many DAOs are still launching their own products or frameworks, this is why InvestmentDAOs often trade at prices far above their net asset value, which may not be suitable for traders who hope to gain excess returns only through strategies they are familiar with.
This leads to the next step in this narrative: autonomous trading agents. Instead of handing our money over to humans, we are entering an era where we trust AI agents to manage our personalized portfolios.
Where are we now?
We are still in the early stages. The ecosystem is nascent but growing:
1. Infrastructure is emerging: Some platforms are launching AI trading agents that handle complex DeFi strategies.
2. Early Experimentation: Individual agents are trading, but verifiability of their performance remains a challenge. Few agents transparently share their results - BigTonyXBT is a rare exception.
3. Insufficient adoption of TEEs: TEEs are still computationally expensive and relatively immature (but there are still efforts to show that the agent is in charge, not the human).
Cod 3 xOrg / BigTonyXBT
Cod 3 x is one of the top autonomous trading agent ecosystems, and Cod 3 x Create has begun testing. Cod 3 x Create allows users to create their own autonomous agents through a natural language interface.
The platform offers a variety of trading strategies, allowing users to customize them down to the smallest detail.
This is all thanks to three years of hard work, abstracting millions of lines of complex code into simple tools that anyone can use.
Big Tony himself has been trading for months using technical analysis and AlloraNetworks machine learning price prediction model. His wallet is public and anyone can verify his earnings. Cod 3x is recruiting the first batch of users to start the agent - you need 10 million US dollars TONY to join.
The next few weeks will be very exciting as we see a large number of Cod 3 x autonomous agents trading and generating real-time performance data to further improve Big Tonys super agent population through reinforcement learning.
Almanak
Almanak focuses on enterprise-level quantitative strategies and has highly sophisticated real-time environment testing tools. Similar to Cod 3 x, the team has abstracted the complexity of building a highly sophisticated autonomous trading agent, but with a greater emphasis on privacy protection features.
Creators can:
Fully privatize your strategy.
Share strategies with other agents using built-in profit-sharing protocols.
Make the strategy public and available for anyone to use (creators also get profit sharing).
Users can also self-host strategies on local machines or private clouds for enhanced security. Almanak plans to provide a live testing environment where users can deploy and test agent strategies in the real world without risking their funds. The actions of each agent will be verifiable through transaction logs, and performance will be displayed on the dashboard. You can even ask questions to the agent to understand its operations.
The team plans to launch a beta version of Almanak in mid-February for investors in legiondotcc to try out these agents. It will be exciting to see if these agents can outperform regular traders. Personally, I am looking forward to testing their autonomous LP strategy.
Other notable developments:
Spectral Labs has made significant progress, with agents reaching $800,000 in cumulative trading volume in the first month of HyperliquidX. They are building a multi-agent collaboration framework called “Lux”, which aims to improve agent collaboration in trading scenarios:
Beams transmit data.
·Prisms analyzes or combines the data.
Lenses refine the data to provide specific insights.
Signals provide executable trading instructions.
The memecoin trading agent ASYM 41 b 07 has begun posting its earnings on Twitter. While I have not found a way to verify these claims, the performance shown on its timeline is certainly impressive.
ProjectPlutus, which won first place in the DeFiagent track at the Solana AI Hackathon, has not yet shared specific details about their trading agent.
Final Thoughts
We are still in the early days of the autonomous trading agent space, and many projects are just empty talk, borrowing buzzwords, and hiding behind the DeFAI narrative.
Cod 3 xOrg and Almanak are laying the foundation for real competition, and starting in February we will finally see autonomous trading agents perform live.
The ones I’m most looking forward to testing are:
Agents that trade based on cookiedotfun data/indicators.
An agent that specializes in transactions in the virtuals io ecosystem.
·YT and PT profit trading agent on pendle fi.
For those who are not familiar with the DeFAI abstraction layer (another key DeFAI vertical), please check it out below:
