Interpreting DeFai’s four popular sections: Why is it said that even an iron tree can bloom?

"Suddenly, a spring breeze comes, and even an iron tree can blossom." How come so many DeFai projects emerge like magic in such a short time? The standards and frameworks haven't been figured out yet, and a new round of DeFai internal war has begun? Okay, next, I will share from a popular science perspective what is going on with several major categories of DeFai projects?

1) In the past two days, @poopmandefi shared a distribution map of DeFai ecosystem projects, which has been widely circulated in the community. Looking at the comments section, there are still too many related projects that have not been included. Many people will be anxious and worry about missing out on wealth codes one by one, but there is no need to be.

First of all, it should be noted that there are certainly some good AI "new" projects among them, such as: $AIXBT, $BUZZ, #NEUR, #GRIFT, #Cod3x, etc., but most of them are new faces that exude an "old" flavor.

The core reason is that most of them are old projects that have been given new expectations through the new narrative of AI Agent, some old projects that have done a lot of work to optimize the experience in the DeFi field but no one cares about them, and some old projects that are difficult to discover in the context of the last round of VC attracting retail investors' attention.

2) There are four major categories in the figure below. I will try to dissect my understanding one by one:

1. AI Abstraction: As the name implies, it is some information processing capabilities of AI large models that are encapsulated in the front-end product experience with which users can directly interact semantically through abstract capabilities. Users can enter some prompts to directly call the transaction interface in the front end of the dialog box, and the AI ​​backend will automatically complete the transaction directly.

Because of this, this type of product is looked down upon at first glance, because there is a lot of friction in the early product interaction experience. For example, the "ambiguity" of the user input prompt and the "precision" of the AIGC backend processing information and executing requirements require a "fault-tolerant" mechanism. Either the user feels that the instructions that can be entered and executed are too simple to compete with the current DeFi experience, or the user enters too many high-expectation instructions and finds that the program backend does not have an accurate execution solver and cannot handle it.

However, this type of product can also gain the trust of a large number of users with its novel interactive mode and solutions to some basic Swap, Staking and other issues. The reason is that its potential is highly predictable. Because the user can input prompts in convenient ways such as text and audio, which are in line with user habits, it will greatly reduce the threshold for use. At the same time, the processing power of the AIGC background will gradually encapsulate more new Solver execution solutions to improve the user experience.

Anyway, this is an attempt to explore a new trading paradigm, just like when Uniswap entered the market with the AMM Swap trading pool paradigm, it was also complained about slippage friction at the beginning. The AI ​​Abstraction segmentation track is indeed tasteless in the short term, but the large paradigm shift opportunities that have been nurtured in the long term are worth paying attention to.

2. Autonomous Portfolio Management & Yield Optimization: This type of product is the result of the last round of DeFi market involution. A large number of projects that want to get a share of the DeFi track have continued to work from the perspectives of personalization, customization, vertical segmentation, and specialized experience, but before they can reap the fruits of victory, the DeFi industry is almost desolate.

Most of these DeFi yield optimization strategies come from the team's ability to monitor and analyze on-chain data, such as transaction depth, capital flow, APY fluctuations, slippage estimates, price deviations, arbitrage space, risk warnings, etc. Based on these real-time on-chain data analysis, a set of execution strategies are formulated, such as position capital allocation, arbitrage opportunity capture execution, Yield income estimation, single pool or portfolio strategy, impermanent loss management, liquidation risk control, etc.

Simply put, the core of this type of product is real-time on-chain data + trading opportunity capture capabilities, plus a complete set of automated analysis and execution experience upgrade capabilities based on smart contracts. At first glance, what does it have to do with AI? The combination point is that data analysis and strategy formulation can be used in the strategy training and fine-tuning of traders to run a set of possible investment opportunities that are more efficient than manual.

Moreover, after combining with AI Agent, the imagination space is even greater. Everyone can use their own strategy to fine-tune an AI Agent with personal trading preferences to automatically help themselves find opportunities on the chain and automatically execute transactions. Making AI Agent a high-level trading assistant for people is a long-term sexy and online narrative. 3. Market Analysis or prediction: This type of product, as a powerful single AI, has captured most of the user's mindshare. For example, @aixbt_agent has indeed become a key information acquisition platform for many traders as a top KOL. However, everyone recognizes the actual application scenario capabilities of AI Agents that only provide trading strategy analysis, but lacks long-term imagination. For example, can my AI Agent monitor the news of AIXBT and automatically help me decide to buy the bottom and arbitrage? And so on. In theory, it is naturally feasible. In fact, AI Agents such as AIXBT are completely capable of autonomously managing user assets, and at the same time, based on their own information decisions, help users to trade, but this step has not yet been taken. At present, the speed of occupying the user's mind of this type of product is so fast, and the commercial realization ability behind it driven by traffic is actually not small.

4. DeFai infrastructure or Platform: This type of protocol covers a wide range. In addition to emerging AI Agent Native platforms such as #ai16z and #Virtaual, other projects such as Bittensor, io, Atheir, @hyperbolic_labs, Vana, SaharaAI, etc. that are related to AI computing power, data, fine-tuning and other business frameworks can be extended here.

After all, AI Agent needs to operate normally, data is oil, computing power is the power grid, reasoning is the transformer, AI Agent is the terminal, etc., they are all upstream service providers.

So, there is not much to say. AI Agent needs to accumulate strength in the second half, and this type of DeFi platform will definitely make great efforts. Originally, AI Agent narrative is only the earliest link in AI Narrative. Frameworks and standards, DeFai, Gamfai, MetAiverse and other focus narratives are inseparable from these AI infra platforms.

That’s all.

Although I have given everyone a clear understanding of DeFai, it does not mean that I am not optimistic. Compared with the current chaotic and difficult-to-judge framework and standard "chaos era" market, DeFai at least replaces a more AI Agent application, and can see progress and expectations step by step through experience, PMF product landing, etc.

This is also a manifestation of the current AI Agent market's popularity from virtual to real. Moreover, so many old species can't find opportunities in the old DeFi era. In the face of new trends, isn't it their opportunity to release their potential?