Viewpoint: Significant pullback, AI Agent narrative evolution still on a smooth path

Author: Haotian

I believe that many people have been confused by the recent market downturn, to the point where the originally vibrant AI Agent new narrative has also been clouded. Thousands of words have converged into a question: Is the AI Agent’s new narrative over? Honestly, I don’t know if the currency price has bottomed out, but the narrative evolution of the AI Agent is still smooth. Next, let’s discuss a few points of view:

Not much to say about comforting words. Like everyone else, those who have high hopes and confidence in AI Agent are enduring the agony of continuous shrinkage in portfolio value during this wave.

The core reason is that although the primary market Build has a clear direction of “value investment research” in chasing AI infra, framework standards, applications, etc., the secondary market is too MEME-ized and the team is somewhat inexperienced. The overall risk in the chaotic AI Agent market is far greater than the opportunity.

So when faced with macroeconomic uncertainty and unclear direction of the big pie, AI Agent race, as a new chain track, naturally became the first object of capital hedging. This makes the track, which already has insufficient liquidity, even more difficult to resist the ability of big market fluctuations. Therefore, small cuts seem to be the norm, and large ones are almost zero. But don’t forget that perhaps because of this, once the market environment stabilizes, the first race to take off is probably the AI Agent race.

2. There are signs that the potential of the new narrative of AI Agents is just beginning to emerge and cannot be allowed to wane. On the one hand, the emerging forces of AI entrepreneurship that overflow from web2 are paying increasing attention to the potential opportunities in the web3 field, and there will be more excellent teams and projects pouring into web3 to explore new possibilities. On the other hand, most of the web3 narratives that are in a deadlock will firmly grasp the new narrative of AI Agents and attempt new breakthroughs, with many well-known VCs and regular army teams in web3 planning to nurture new projects.

In short, AI Agent represents the hope of the entire village, and its potential has just begun to explode. It does not have the fundamental aspect of announcing the end of suspension now. Especially the collective entry of some VCs and mature project parties will shuffle the chaos caused by the team being too informal, and further improve the industry’s technical threshold (standardization) and internal competition. For retail investors, it means that the stability and maturity of the track will be improved.

3. #ai16z 和 #Virtual The two major AI Agent camps have exposed their own problems after this wave of turmoil, including the confusion in resource integration on the side of ai16z, and the uncertainty of the post-development of the Virtual closed-source ecosystem. After this big wave of turbulence, being able to detach from the Fomo state and return to calm is also a good thing. It is not conducive to the subsequent outbreak of greater potential for any force to be excessively mythologized.

After all, as a big business IP, ai16z is like a spiritual totem. Open source innovation and community vitality represent the true future of ai16z. No matter how market Fud @shawmakesmagic’s personal cultivation is, as long as the developers behind the scenes still revolve around the ElizaOS framework, the future of ai16z is immeasurable;

The drawback of Virtual’s closed-source ecology lies in the positive/negative spiral characteristics of its Tokenomics, which is outstanding in both the market upswing and the pullback period, but behind this is the explicit fact that its hype attribute is greater than the application value, and I am more concerned about how the Virtual closed-source model can create more @aixbt_agent monolithic AI applications with a higher level of experience.

4. AI Agent has hastily gone through a round of AI MEME, AI standalone applications, AI frameworks and standards, AI chain engineering, AI metaverse, DeFai, etc. in the past three months. This kind of evolution speed is actually very dangerous. Taking the framework and standards as an example, no matter how exaggerated it is, it cannot change the suspicion of “drawing pie”. At present, the track is still in the stage of using standalone AI to change user Mindshare, without the richness of the application market, only internal framework and standards, which is meaningless.

Ultimately, the market’s confusion about the value assessment system of the new direction of AI Agent is the key issue. Many people tend to assign higher value to seemingly elusive and ethereal directions such as framework standards and metaverse, but in reality, it’s not the case.

The foundation of the development of the AI Agent track is “application-oriented”. Any concept and direction that cannot be built on the basis of application is of no avail. If it is found to be a mess after a hasty round of speculation, there is no doubt that the focus of the next stage of competition is the individual AI application. Multimodal interaction, AI metaverse, and DeFai can all be hyped, but please take the Agent out for a walk first.

5. The overall trend of “chaining” of AI agents will still exist, but it is time for phase engineering delivery. Currently, the ElizaOS framework has only some practical primary solutions in the directions of Memory, Character, etc., and there will be many difficulties to overcome in incorporating the decentralized thinking of chaining.

For example, DA capability. Traditional layer2 achieves limited DA extension through off-chain zk proof certification and on-chain distributed verification. However, how to ZK the dynamic prompt input and output of LLMs large models, and how to verify data availability after a long period of time are big challenges. Another example is Oracle capability. Feeding AI Agent with on-chain + off-chain data for implementation seems simple, but balancing real-time performance and computational costs, as well as the complexity of on-chain data and the trustworthiness of off-chain data, are all major challenges.

So, I still firmly believe that ‘chaining’ will be the main theme of the next evolution of AI Agents, but the key is how to deliver some practical, scalable, and verifiable solutions. Simply applying the narrative of the old chain to AI Agents is not impossible, but there are many detailed engineering implementation details that need to be refined.

Above.

Although there is a high probability that the AI Agent direction will continue to be dominated by ‘chaos’ in the future, the AI Agent is a new direction for redeeming old narratives, which means that the market will be extremely picky and scrutinize all its evolutionary trends.

Like some who try to bring VC coins like FDV into the AI Agent field, like AICC, which is full of elite plunderers, like some Li Kui gangs who hide evil intentions and only treat web3 as a financing channel. I would like to advise that the road to the counterattack of AI Agent is ultimately the grassroots power. It is built and defeated by the community. Projects that do not bind the interests of the community and the masses cannot go far.