AI Summer’s development model is similar to DeFi, from popular applications to infrastructure, and from imitators to innovators. If we discuss from upstream to downstream, it is computing power + data ($IO $HEU $GRASS) -> model ($TAO) -> application (such as AI Agent), and there will be corresponding frameworks at each level.

I would like to add that there are many directions for applications, and AI Agent is one of the directions we are currently hyping. Computing power + data and applications are what I talk about more. I plan to talk more about things I haven’t talked about before, which can be regarded as pushing me to take the initiative to learn.

At the time of FLock TGE (also Bybit’s New Year’s Eve Launchpool), let’s talk to you about the @flock_io project.

To understand simply, FLock is an AI model training middleware, between computing power + data providers and AI application developers. Developers provide decentralized/modular computing power + privacy-protected data to help them conduct better and more decentralized AI model training.

The knowledge points involved here are "decentralized/modular computing power" and "decentralized storage of data + privacy protection".

Friends who have read my previous tweets should know the reason why I am optimistic about $HEU: the AI ​​Agent framework launched by Heurist is actually looking for a buyer's market for its own computing power. . I like its self-produced and self-sold model.

As long as the development of AI-related applications will involve computing power, data and training. FLock is equivalent to solving all the troublesome things that developers need to deal with through its own infrastructure - developers only need to adjust the training model, and the computing power and data are all solved by FLock.

The above part is mainly to talk to you about the ecological niche of FLock.

Secondly, when making products, you must use a combination of punches.

FLock’s combo punches are AI Arena, FLAlliance and AI Marketplace trio.

AI Arena is an AI training platform where AI developers come here to train AI and adjust models.

FL Alliance is a collaboration platform for data providers and AI developers. Data providers can provide privacy-protected proprietary data to AI developers. to further optimize the AI ​​model.

AI Marketplace is a model hosting and trading platform. Simply put, the AI ​​Marketplace is a platform that brings together a bunch of people to build, improve, and use AI models.

Using the worldview of martial arts as an analogy, AI Arena is a place where faces are pinched and basic values ​​are added, FL Alliance is a place where you can upgrade and learn skills and moves, and AI Marketplace is A gang, a place where everyone comes together to discuss martial arts and learn from each other.

Through the above three-piece set, FLock’s ultimate goal is to produce more and more high-quality AI models.

Currently, the total number of FLock tokens is 1 billion, and the initial circulation is 12.88%, as shown in the figure below. It will airdrop tokens to all participants of previous Flock products and testnets. Investors and teams are locked in for 1 year.

Next let’s take a look at FLock’s investment background:

Faction, the lead investor, has invested before Lens, Puffer these protocols.

Well-known investors include Animoca Brands, DCG, OKX Ventures, GSR ($USUAL $MOVE market maker). DCG is the parent company of Grayscale. Currently, Grayscale's focus is also moving closer to AI. For example, Grayscale Research has just updated the top 20 tokens in Q1 of 2025 and added two new AI ecological tokens, $VIRTUAL $GRASS.

It is worth mentioning that FLock has also received funding from the Ethereum Foundation Academic Grants Round 2024.

Finally, let me express my opinion: I think the model of small balls driving big balls is very suitable for the AI ​​Agent track, such as AIXBT/VIRTUAL, SPORE/ PHA. To predict whether an AI Agent infrastructure can take off, it is necessary to observe the performance of its ecological Agent, that is, to drive the continued growth of the parent project through the mindshare of the sub-project occupying the market. In the same way, we can also apply this methodology to FLock - formulate investment plans by observing the performance of FLock ecological sub-projects.