DeFAI New Era Four Major Tracks: AI Abstraction, Automatic Investment Management, Market Prediction, Infrastructure

DeFai has at least substituted a more AI Agent application, which can see progress and expectations step by step through experience, PMF product landing, and so on. This article is based on an article written by Haotian and compiled by PANews. (Summary: AI Agent Market Dynamic Analysis: DeFAI, Game Agents and Investment DAOs Become New Hot Spots) (Background Supplement: A Quick Look at the Four Categories of the AI Agent Track: Framework, Launchpad, Application and Meme) Suddenly like a spring breeze overnight, iron trees can also bloom, how can so many DeFai projects emerge in such a short time like magic? The standards and frameworks have not yet been clarified, and a new round of DeFai involution wars has begun? Well, next, let me share from the perspective of popular science, what is going on in several major categories of projects in DeFai? 1) In the past two days, Poopman has shared a map of DeFai ecosystem projects, which has been widely disseminated in the community, and there are still too many related projects in the comment area that have not yet been included. Many people will be very anxious, worried about missing a wealth password, in fact, it is not necessary. First of all, it should be noted that of course, there are some good AI "new" projects of Mindshare at present, such as: $AIXBT, $BUZZ, #NEUR 、#GRIFT, #Cod3x, etc., but most of them are new faces that exude an "old" flavor through and through. The core reason is that most of them are old projects that have been re-given new expectations through the new narrative of AI agents, some old projects that have done a lot of optimization experience work in the field of Decentralized Finance but no one cares, and some old projects that are difficult to be discovered in the context of the last round of VC traction retail investor. 2) There are four major categories in the figure below, I try to dissect my understanding one by one: 1, AI Abstraction (AI abstraction): As the name implies, some through abstraction capabilities, the information processing capabilities of AI large models are encapsulated in the front-end product experience where users can directly interact semantically, and users can enter some prompts can directly call the transaction interface in the front-end of the dialog box, and the AI back-end automatically completes the transaction directly. Because of this, such products look white at a glance, because the early product interaction experience has a lot of friction, for example, the "ambiguity" of the user input prompt and the "precision" required by the AIGC back-end to process information and execute the required "fault-tolerant" mechanism, or the user feels that the instructions that can be entered and executed are too simple to counter the current Decentralized Finance experience, or the user enters too many high-expectation instructions. It was found that the program background did not accurately execute Solver and could not be handled. However, such products can also gain the trust of mass users with novel interaction modes and some basic Swap, Staking and other problems. The reason is that its stamina is highly predictable. Because the way users input Prompt can be a convenient way to use text, audio, etc., it will greatly drop the threshold of use, and the processing power of the AIGC backend will gradually encapsulate more new Solver execution solutions to improve the user experience. Anyway, this is an attempt to explore the formalization of new transactions, just as Uniswap brought the AMM Swap trading pool to the market when it was regularized, and at first it was complained that Slippage was frictionless. The AI Abstraction subdivision track is indeed very chicken in the short term, but the long-term gestation of the large regularization transfer opportunity is worth following. 2, Autonomous Portfolio Management & Yield Optimization (Automated Portfolio Management and Yield Optimization): This type of product is the result of the last round of Decentralized Finance market involvement, a large number of projects that want to get a piece of the Decentralized Finance track from personalization, customization, vertical segmentation, Specialized experience and other angles continue to cultivate, but before the time to harvest the fruits of victory, the Decentralized Finance industry is almost desolate. Most of the strategies for this kind of Decentralized Finance revenue optimization come from the team's monitoring and analysis capabilities of on-chain data, such as transaction depth, capital flow, APY Fluctuation, Slippage estimate, price deviation, Arbitrage space, risk warning, etc., based on these real-time on-chain data analysis, and then develop a trap execution strategy. For example, Position fund allocation, Arbitrage opportunity capture execution, yield yield estimates, single pool or portfolio strategies, Impermanent Loss management, liquidation risk control, and so on. Simply put, the core of such products is real-time on-chain data + transaction opportunity capture capabilities, plus a whole trap based on smart contract optimization of automated analysis, execution operation experience upgrade capabilities. At first glance, what does it have to do with AI? The point of integration lies in data analysis and strategy formulation, etc., which can train and fine-tune the trader's strategy, and run a trap more efficient than the human possibility of investment opportunities. Moreover, after combining the AI Agent, there is more room for imagination, and everyone can use their own strategy to fine-tune an AI Agent with personal trading preferences to automatically help them find opportunities in the on-chain and automate the execution of transactions. Making AI agents people's high-end trading assistants is a long-term sexy and online narrative. 3, market analysis or prediction: This kind of product as a super single AI has captured most of the users Mindshare, for example, aixbt has indeed become a key information acquisition platform for many traders as a top KOL. However, everyone recognizes the ability of AI Agent that only provides trading strategy analysis for its practical application scenarios, but there is less room for long-term imagination. For example, can my AI agent monitor AIXBT messages and automatically help me decide to catch the bottom and Arbitrage? And so on. In theory, it is naturally feasible, in fact, it is entirely possible for AI agents such as AIXBT to independently host user assets, and at the same time make decisions based on their own information to help users conduct trading operations, but this move has not yet come out. Only at present, the mental occupation speed of users of such products is so fast, and then the business monetization ability behind the traffic drive, the imagination space is actually not small. 4, DeFai infrastructure or platform (infrastructure and platform): This type of protocol can cover a wide range, in addition to #ai16z 、#Virtaual and other emerging platforms such as AI Agent Native, other such as Bittensor, io, Atheir, Hyperbolic, Vana, SaharaAI and so on Computing power, data, fine-tuning and other business-related project business frameworks can be extended here. After all, AI agents need to function properly, data is oil, computing power is the power grid, inference is transformers, AI agents are terminals, and so on, all of which are upstream service providers. Therefore, there is not much to say, the second half of the AI agent must be accumulated, and this kind of decentralized finance platform will definitely make a big effort. Originally, the AI Agent narrative also...