Although it is highly likely that the AI Agent direction will continue to be characterized by "chaos" in the near future, AI Agents are the new direction meant to redeem old narratives. This means that the market will be extremely picky and scrutinize every evolutionary trend of AI Agents with great rigor.

Many people have been bewildered by the recent market downturns, to the extent that the originally vibrant new narrative of AI Agents has also been overshadowed. All words and deeds converge into a single question: Is the new narrative of AI Agents over? To be honest, I don't know if the currency prices have bottomed out, but the evolution of the AI Agent narrative still has a clear path ahead. Here are a few perspectives:

1. Comforting Words Are Few: Like everyone else, those who have high expectations for AI Agents and hold confidence in them are enduring the agony of continuous value shrinkage in their positions.

   The core reason is that while the primary market has clear "value research" directions in pursuing AI infrastructure, framework standards, and applications, the secondary market is overly MEME-ized, and the teams are somewhat makeshift. The risks in the overall AI Agent market amidst the chaos still far outweigh the opportunities.

   Therefore, when faced with macroeconomic uncertainty and unclear directions in the broader market, the AI Agent track, as a new and sharp track on the chain, naturally becomes the first target for capital to hedge risks and liquidate. This makes the track, which already has insufficient liquidity, even less capable of withstanding major market fluctuations. As a result, it seems inevitable that the track will experience significant declines, or even near-zero values. But remember, perhaps precisely because of this, once the overall market stabilizes and opportunities re-emerge, the AI Agent track is likely to be the first to take off.

2. Signs Indicate the Momentum of the New AI Agent Narrative Has Just Begun: On one hand, the emerging forces of AI entrepreneurship spilling over from Web2 are increasingly noticing the potential opportunities in the Web3 domain. More excellent teams and projects will flood into Web3 to explore new possibilities. On the other hand, the majority of the old narratives in Web3 that are stuck in困境 will seize the new narrative of AI Agents to attempt new breakthroughs. Many well-known VCs and established Web3 teams are currently planning and incubating new projects.

   In summary, AI Agents represent the hope of the entire village, and the momentum has just begun to erupt. There is no fundamental basis for it to shut down and declare an end now. In particular, the collective entry of some VCs and mature project parties will reshuffle the current chaos caused by makeshift teams and further raise the industry's technical threshold (standardization) and internal competition. For retail investors, this means that the stability and maturity of the track will be improved.

3. The Two Major AI Agent Camps, #ai16z and #Virtual, Have Both Exposed Their Own Issues After This Turbulence: For instance, #ai16z has shown confusion in resource integration, and #Virtual's closed-source ecosystem development has uncertainties in its sustainability. It is also a good thing to be able to detach from the FOMO state and return to calmness after such a major fluctuation. Over-mythologizing any one force is not conducive to the subsequent larger momentum explosion.

   After all, as a major commercial IP, #ai16z is originally a spiritual totem-like figure. Open-source innovation and community vitality represent the true future of #ai16z. No matter how the market FUD @shawmakesmagic's personal cultivation is, as long as the developer community continues to revolve around the ElizaOS framework, the future of #ai16z is immeasurable.

   The disadvantage of #Virtual's closed-source ecosystem lies in the positive/negative spiral characteristics of its Tokenomics, which are prominently displayed in both rising and回调 periods. This reflects that its speculative nature outweighs its application value. I am more concerned about how #Virtual's closed-source model can create more individual AI applications with a high level of experience like @aixbt_agent.

4. AI Agents Have Hastily Gone Through a Round of Evolution in the Past Three Months in Directions Such as AI MEME, Individual AI Applications, AI Frameworks and Standards, AI Chain Engineering, AI Metaverse, DeFai, Etc. This kind of evolutionary speed is actually very dangerous. Take frameworks and standards as an example; no matter how eloquently they are described, they cannot change the suspicion of "pie in the sky." The track is currently in the stage of using individual AI to change user mindshare. Without a rich application market, it is meaningless to just compete on frameworks and standards.

   Ultimately, the market's value assessment system for the new direction of AI Agents is in disarray. Many people believe that the more seemingly intangible directions like frameworks and standards, metaverse, etc., can be valued higher, but this is not the case.

   The development foundation of the AI Agent track is "application first." Any concept and direction that cannot be based on applications is useless. If, after a hasty round of speculation, it turns out to be a mess, then undoubtedly, individual AI applications will be the focus of the next stage of competition. Multimodal interaction, AI metaverse, DeFai can all be hyped, but please first show off the Agent.

5. The Overall "Chainization" Trend of AI Agents Will Continue, But It Is Time for阶段性工程交付. Currently, the ElizaOS framework only has some primary practical solutions in directions such as Memory and Character. To incorporate the decentralized ideas of chainization, there will be many difficulties to overcome.

   For example, DA (Data Availability) capabilities: Traditional Layer 2 achieves limited DA expansion through off-chain zk-proof and on-chain distributed verification. However, how to ZK-ize the dynamic prompt inputs and outputs of LLMs (Large Language Models) and how to verify data availability in the long term are major projects. Another example is Oracle capabilities: It seems simple to feed AI Agents with real-time on-chain and off-chain data, but balancing real-time performance and computational overhead, as well as dealing with the complexity of on-chain data and the trustworthiness of off-chain data, are all significant challenges.

   Therefore, I still firmly believe that "chainization" will be the main theme of the next evolution of AI Agents. However, the key is to deliver some truly valuable, scalable, and verifiable solutions. Simply applying the old chain narratives to AI Agents is not impossible, but there are still many detailed engineering implementation details that need to be refined.

In Conclusion

Although it is highly likely that the AI Agent direction will continue to be characterized by "chaos" in the near future, AI Agents are the new direction meant to redeem old narratives. This means that the market will be extremely picky and scrutinize every evolutionary trend of AI Agents with great rigor.

For those who try to bring high FDV VC coins into the AI Agent field, those who are filled with elitist exploitation like AICC, and those who harbor ill intentions and only see Web3 as a financing channel, let me offer a piece of advice: AI Agents are ultimately the path for the underdog to rise. Success lies in the community, and failure also lies in the community. Projects that do not bind the interests of the community and build together will not go far.