AI Agent: Opening a New Chapter in the Era of Intelligence

Time flies, and before we know it, it's already 2025. AI technology has long been ingrained in people's hearts, but many still have misconceptions about the accurate definition of Agent( intelligent agent).

In fact, for a project to be called an "AI Agent", its problem-solving logic must follow the steps below:

1. Perception
2. Think
3. Action

Taking a common AI automatic investment cryptocurrency project as an example, its working principle is as follows:

1. Awareness: Get the latest cryptocurrency news.
2. Consideration: Analyze the impact of news on specific cryptocurrencies
3. Action: Decide to buy or sell based on the analysis results.

This process is very intuitive and easy to understand.

So, is the well-known ChatGPT considered an AI Agent?

Although ChatGPT considers itself an AI Agent, it admits that this is only a broad definition. If we define "AI Agent" broadly enough that it only needs to possess perception, decision-making, and action loops (, then ChatGPT can be seen as an AI Agent. However, if the requirements for an AI Agent include richer perception and physical/system-level execution capabilities, then ChatGPT is more like a conversational intelligent component, rather than an Agent with complete multimodal or physical environment operation abilities.

Why is there this distinction? There is a subtle linguistic difference here.

Imagine that when you want to use ChatGPT, you simply say "I want to use AI" without adding another word to say "I want to use AI Agent."

Therefore, when we specifically mention the term "AI Agent", we must be emphasizing certain specific characteristics.

We can further summarize the characteristics of AI Agent:

• More focused on solving problems in specific areas
• Need to obtain more data from the physical world

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-99e6e198e6f4c6c0a1fb7773baaf314d.webp(

Solving Vertical Domain Issues

Although in the future AGI era, GPT may only be limited to generating dialogue AI, which is a typical vertical application. However, in today's 2025, GPT) certainly also includes competitors like Gemini, Grok, etc. (, which have already become all-round players of this era.

Therefore, from a definitional perspective, the current AI Agent needs to be more focused on vertical domains than GPT.

Obtain More Physical World Data

AI Agent needs to have the ability to actively acquire knowledge of the physical world.

For example, another well-known AI Agent, which is currently the top trending AIXBT) on social media, Mindshare ranked first (, has undoubtedly acquired more data from the physical world, such as various cryptocurrency news. This allows this purple frog-themed Agent to make various comments and predictions.

What impressed me the most was that AIXBT predicted that a certain cryptocurrency would be listed on a trading platform within 6 hours, and it really happened that night.

In summary, we can derive the complete definition of AI Agent:

1. Adopting the "Perceive-Analyze-Act" processing flow
2. Focus on solving problems in specific areas
3. Utilize more data from the real physical world

Only by understanding the definition can we further discuss.

So, what exactly can the AI Agent do?

This question is actually a variant of "What can cryptocurrency do?" and is a question that many outside commentators on cryptocurrency have been questioning.

After trying dozens of Agents across multiple frameworks, I have summarized them into the following main categories:

![AI Agent Track Vision: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-ee64b5cc77cf84fba92e3948e8ca838c.webp(

1. AI Investment

This type of application is quite common. Users can fund the AI Agent wallet, and then the AI Agent analyzes market trends based on real-time news and autonomously makes buy or sell decisions.

Comment: The concept is interesting, but the wallet's security is questionable.

2. AI Virtual Companion

These applications are essentially what people commonly refer to as GPT shells. However, compared to the past, they have incorporated some personality limitations.

For example, Eliza repeatedly emphasizes that she is a real girl, not an AI. Similarly, AVA, a white-haired beautiful girl character, is skilled at creating videos and analyzing market trends through video.

Comment: In real life, there is significant pressure for political correctness, and it has been proven that people still prefer idealized images.

3. Opinion Output Type

This type of Agent focuses on commenting on specific areas.

For example, AIXBT comments on cryptocurrencies and even conducts some comparative studies. Similarly, Zerebro, although its concept is rather fantastical, gives an overall impression of a mad artist who publishes bizarre theories every day. There is also an AI Agent that mimics a certain well-known effective accelerationism )E/ACC( critic, which releases some peculiar statements in a similar speaking style.

Comment: Although many people question the existence and significance of such applications, in today's era, as long as there is traffic, there is significance.

4. Virtual Idol Category

Creating virtual idols based on AI can release music works, such as Luna.

5. Prediction Category

There are projects on a certain platform that specialize in predictive market forecasting, which can also be seen as a branch of opinion.

6. AI Resource Sharing Class

For example, FXN based on the Eliza framework aims to provide a bundled integrated service for users who are unwilling to purchase various AI memberships separately, paying per use to save costs.

Comment: It can be understood as an AI version of group buying services.

7. Other Categories

Including painting, GIF generation, music creation, etc., which will not be listed one by one here. In general, they do not exceed the scope of the current AI industry multimodal large models.

So, what is the AI framework?

The current AI Agent field is flourishing, which is quite similar to the ICO boom of previous years and the DeFi Summer.

If we make an analogy, then platforms like AI16Z, Virtual, AVA, etc. are equivalent to public chains, and using their frameworks, one can quickly launch personal AI Agents.

Of course, using these frameworks requires payment.

For example, if you want to launch an AI Agent using AI16Z/Eliza, you need to allocate a certain percentage of tokens to the AI16Z treasury and stake AI16Z coins. This also explains why the funds in the AI16Z treasury are continuously increasing.

To launch an AI Agent using the Virtual framework, you need to pay 100 Virtual coins and pair your tokens with Virtual to form a liquidity pool, which is similar to the model of certain public chain ecosystems.

The overall model of AVA and Swarms is similar to the previous two, but due to their later launch, their ecosystem is relatively less rich; however, it is also actively developing.

It is worth mentioning that AVA originated from an AI project called Holoworld supported by an incubator, and has decisively transformed into an AI Agent, focusing on video generation.

The founder of Swarms is depicted as a genius youth. This framework has been operating in traditional fields for many years, with an overall emphasis on technology, highlighting that tasks can be accomplished through the collaboration of multiple agents. This is also the origin of the name "Swarm".

Of course, there is also a piece of gossip news, which is that the founder of Swarms was severely criticized by Shaw of AI16Z for having poor technology.

However, speaking of which.

Applications are applications, and tokens are tokens.

The most important feature of these frameworks, or the biggest difference from traditional AI frameworks, is that they can conveniently help developers issue tokens.

Or more bluntly: essentially these AI frameworks are also heavily modified based on the existing AI industry framework, standing on the shoulders of giants, with their greatest originality lying in the token issuance module.

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-79f26a6af6eae60267e25a6b5753afa2.webp(

Imaginary AI Developer: Founding employees of top AI companies, PhDs from prestigious universities, earning a salary of one million dollars per year.

Real-world AI developers: issue coins first and then talk.

Therefore, using these frameworks, if you want to issue an AI Agent, the process is roughly as follows:

1. Determine the theme
2. Choose the appropriate framework and make payment
3. Issue tokens and list them
4. Create social media accounts for the AI Agent to publish content.
5. Add more features to the AI Agent ) needs to be developed (
6. Develop the official website, establish a community, contact exchanges
7. Establish your own AI framework system
8. Encourage more people to identify new themes...

This process is quite reminiscent of "A Record of a Mortal's Journey to Immortality."

The corresponding valuation is approximately ) for reference only (:

1 100 10K 100K 1M 10M 100M 1B

In contrast, the development process of traditional AI companies:

1. Determine the theme
2. Financing
3. Burn Money
4. Financing
5. Burn Money
6. Financing
7. Listing

It can be seen that the model of cryptocurrency + AI is actually about getting listed first and then talking, which is a literal meaning of "disruptive" innovation.

In addition, due to the outstanding performance of these AI frameworks in the secondary market ), they can reach a market value of 1-2 billion USD without needing to be listed on any large exchanges (, making them currently the only hot track in the cryptocurrency field.

Nowadays, it has reached a point where the streets are empty of people.

It must be acknowledged that there is indeed a overheating phenomenon in the short term. If there were an AI Agent fear and greed index, the current greed index would be at least 90, and it might have even exceeded the limits.

We often see some genuine AI developers from outside the circle criticizing these projects for lacking technical content.

However, my point of view is that the best use case for cryptocurrency is "quickly completing market discovery," although this may lead to some bubbles.

Excellent products and founders can quickly gain exposure and community support, while the founders of inferior products will immediately face negative validation from the market.

In this fierce survival of the fittest, it is possible that innovative products that traditional AI developers did not expect could emerge.

Let us look forward to the arrival of that day.

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-a79e8490e7ed0db21134773e3999d126.webp(

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-3320e1779b06652e2b46806c1822b550.webp(

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-759b2548cfa5616ab5f51e7a169fd191.webp(

![AI Agent Track Imagination: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-92a398c7607e83847f0a4f0913322818.webp(

![AI Agent Track Vision: Do Androids Dream of Electric Sheep?])https://img-cdn.gateio.im/webp-social/moments-549230279e40b9377a0c043cf8e93b2b.webp(