Following the main line of AI Agent, how to select potential Tokens on CEX?

CEX also has a trap logic to pick treasures along with the trend.

I believe that many people feel like they are picking up treasures everywhere when they open DEX, after all, there are popular projects with a size of 10-300M and good technical quality everywhere. However, when they open CEX, they are surrounded by old narrative projects with a size of 300M-3B, and their hearts are filled with mixed feelings. But in fact, CEX also has a logic of picking up treasures along the trend, let me explain briefly:

single AI application

Pure MEME doesn’t need much explanation, it has always been about creating hype; Individual AI, due to the need for a good user experience, can dominate in the short term; Launchpad platform will definitely be taken advantage of by Virtual due to its first-mover advantage; Frameworks and standards, due to the complex technical logic behind them, have great potential for imagination, but currently are still in a chaotic stage of development; Only the direction of AI Agent’s blockchainization has a relatively clear path to follow.

2）The logic of “chaining” must not be that a certain public chain jumps out and says, I can support all AI Agents running on the chain. There must be such public chains, for example,

@NEARProtocol, ICP, and BNBChain all have the potential. However, the problem is that at this stage, AI Agent is still in the early stage of MEME-izing ‘issuing assets’, it can only run AI Agent, but cannot stimulate the enthusiasm of developers, which is useless.

So the first wave of “chaining” narrative must come from the ai16z ecosystem series. Such as @focEliza.

3. And the core point of ‘chaining’ is to build infrastructure for better compatibility and service to AI Agent for large language models LLM, such as:

1. To achieve autonomous generation of private keys and asset management, AI Agent must rely on TEE infrastructure. Therefore, projects that can provide mature TEE solutions are potential Alphas in the narrative of CEX. This is because TEE has always been inconspicuous under the competition pressure of traditional cryptographic algorithms such as ZK, MPC, and FHE, but suddenly it has become the core infrastructure of the new wave of AI narratives.

Imagine some small market cap, low visibility, technically sound projects that have stepped onto a new trend. They are definitely potential targets among many CEX targets. (The logic behind the $PHA performance show is exactly like this. If you agree with the value of TEE infrastructure for AI agents, the potential of TEE track is definitely still there.)

2. To achieve the hierarchical optimization and matching of Memory in AI Agent, the goal is to build Data Availability (DA) capability suitable for AI Agent. Traditional EVM public chains have extended many exciting narratives in the DA section, such as Blob extension space, third-party DA War, etc.

How to build the DA capability exclusive to AI Agent will also be a topic of focus: including how the DA block space accurately records the LLM context and effective semantics, how the Character role setting plug-in interacts with the DA block in real time, how data in multiple block spaces can be effectively matched with multimodal interactions, and the cost of storing data in multiple block spaces (short-term, long-term, working memory, etc.). These are all problems that the new DA Builder urgently needs to overcome. This is also the reason why focEliza focuses on DA capability as the core breakthrough point.

3. To achieve the verifiability of single-mode and the trusted interaction of multi-mode, AI Agent relies on the trusted validation processing at the public chain level. While TEE infrastructure solves the generation of large models and the application of private keys, it is obvious that TEE cannot solve the problem of physical blasting of single-point TEE hardware and the consensus verifiability problem of TEE executing programs.

In the long run, all of these rely on the natural decentralization of blockchain nodes for validation consensus and smart contract collaboration and invocation environment to achieve.

In other words, a powerful consensus blockchain distributed system will make the intelligence of AI Agent “reliable” and guaranteed. So following this logic, efforts are currently being made to provide AI Agents with underlying zkVM framework, ZK Oracle oracle solution, ZK Bridge cross-chain solution, and “chain abstraction” public chain-level application solutions, all of which are part of this current trend of “chaining”.

Above

When a series of solutions such as TEE+DA+Oracle+zkVM+ chain abstraction mature to provide the ‘chaining’ capability for AI Agents, at this point, the decentralized distributed computing power, decentralized fine-tuning inference environment, decentralized data sources, decentralized IP communication and incentives, and a series of AI Platform platforms are needed for AI Agents to truly become a ‘must-have’.

To some extent, based on the efforts of the AI Agent chain direction, large AI platform projects such as io, Aethir, Vana, SaharaAI, etc. will have a place to be used.

Clear up this logic, how to get the truth in the chaotic era on top of DEX, how to find treasure in the silence within CEX, no need to say more, there will naturally be a clear judgment.