In summary, a16z has provided a report on AI+Crypto, which abstracts three core directions from 11 specific use cases: 1) Agent infrastructure: mainly solves basic problems such as identity authentication, cross platform collaboration, and human-machine differentiation for AI agents. Including Universal Agent Identity, Proof of Personahood, communication standards and governance rules between agents. It's like giving each AI agent a family ID card, bank card, and social security card, allowing them to legally "work" and "transact" in the digital world. The current agents are isolated islands that operate independently, and this infrastructure aims to connect them into a super network that can collaborate deeply. The programming assistant you have trained in VSCode can seamlessly switch to GitHub Copilot to continue working, equivalent to the Interoperability and Interoperability aggregation layer of the Agent era; 2) Context data layer: mainly solves the persistence problems of AI's "memory" and "knowledge". This includes cross platform context preservation, IP authentication and copyright protection, data traceability, and value distribution. The current AI LLM does not have a memory layer design, or "goldfish memory". Every time you use AI, it's like finding a temporary job, and you have to retrain every time. With this data layer, AI is like having a 'long-term memory', where your professional knowledge, work style, and project experience can all be transformed into storable and tradable digital assets. Equivalent to the Data Availability layer of the Agent era. Imagine that the coding habits of top programmers can be directly "sold" to AI to guide beginners and monetize their knowledge directly; 3) AI native finance: mainly solves the payment settlement and incentive distribution system of AI economic system. Including micro payments between agents, web crawler payment mechanisms, new commercial advertising models, etc. Nowadays, platforms are eating meat, creators are drinking soup, and users are being targeted. AI native finance aims to ensure fair distribution of every link in the value creation chain. The main theme is' Who creates value, who gains benefits'. If one day, AI agents can directly trade services with each other, and users can receive token rewards for contributing data, even if their personal data is used by AI, they can receive corresponding profits. This step is equivalent to the Settlement layer of the Agent era. Overall, the 11 specific cases of a16z cover a wide range: from persistent AI interaction context, agent universal identity, decentralized computing power network (DePIN), to micro payment revenue sharing, IP registration and authentication, crawler payment mechanism, to privacy advertising, AI companionship, and so on. The report also provides a time forecast: PoP identity authentication and DePIN computing power aggregation will be implemented first in the short term, the trading market between agents will emerge in the medium term, and a complete AI autonomous economic ecosystem can only be achieved in the long term. But personally, I am more inclined towards Agent C-end customization in the Bullish Agent Universal Identity+Agent Context Data Layer Federation. After all, the era of truly personal AI assistants only begins when every AI has a lasting memory and cross platform identity. The above three directions are essentially aimed at building a "digital society where AI can autonomously participate in economic activities", so that AI is no longer just a passive tool, but a digital "worker" who can independently create and exchange value. Obviously, now is the time for AI Infra to establish standards and seize the opportunity to land. Whoever can secure a spot first has the chance to reap the biggest dividends.