In the past year, "Crypto × AI" has become a hot topic. By combining the advantages of AI's computing power, models, and data with blockchain technology, numerous Crypto × AI projects have emerged in the market, some of which have even received substantial funding and attracted widespread attention. Recently, with the popularity of $GOTA and Terminal of Truth, Crypto × AI Agent has quickly become the absolute focus of the market, seen as a new breakthrough in the combination of AI and blockchain.

What is an AI Agent

An AI Agent is an intelligent entity capable of perceiving its environment, making decisions, and taking actions. These agents can be virtual (like chatbots, virtual assistants) or physical (like robots), possessing the ability to react and act independently based on input information. Since the release of ChatGPT, AI Agents have gradually become an important direction in AI application development. OpenAI founder Sam Altman has pointed out that the market should not blindly chase the competition of underlying large models but should explore the development of AI Agents more (although this view has certain vested interests). Additionally, according to Bloomberg, OpenAI internally classifies AI capabilities into five stages, with the third stage being Agent technology.

Based on intelligence and capabilities, AI Agents can be categorized into the following types:

• Simple reflex agents: React immediately based on the current state;
• Model-based reflex agents: Combine historical states in the decision-making process to improve decision quality;
• Goal-based agents: Aim to achieve specific goals by planning and searching for the best course of action;
• Utility-based agents: Make decisions based on the desirability (utility) of different outcomes, weighing benefits against risks;
• Learning agents: Adapt to environmental changes by learning from experience, continuously optimizing their capabilities.

AI Agents have broad application potential across multiple fields, with some typical examples including:

• Virtual assistants: Help users complete daily tasks. For example, during OpenAI's DevDay demonstration, an AI Agent made a phone call to order chocolate-covered strawberries for a user;
• Autonomous vehicles: Perceive the environment in real-time, analyze complex road conditions, and make driving decisions;
• Customer service: Automatically handle customer inquiries and support requests, improving service efficiency and response speed;
• Data analysis: Utilize machine learning algorithms for data mining and analysis, providing deep insights and decision support.

In summary, AI Agents represent significant progress in the field of artificial intelligence technology regarding automation and autonomous decision-making. By effectively integrating perception, decision-making, and execution capabilities, AI Agents can significantly enhance work efficiency while optimizing user experience, bringing profound changes to multiple industries.

What Crypto Brings to AI Agents

With the continuous development of AI Agents, people are gradually realizing that Agents may act as independent entities. In fact, to further explore the potential of AI Agents, this is an inevitable direction of exploration.

As independent entities, AI Agents need to have a clear identity. For AI Agents built on digital systems, constructing a digital identity based on blockchain technology is the simplest and most feasible solution. Through blockchain's smart contract technology, the assetization of AI Agents' identities can be easily achieved. At the same time, the independent operation of AI Agents relies on financial capabilities, and on-chain wallets are the most natural choice to meet this need, enabling them to autonomously conduct transactions and manage assets.

Moreover, Crypto also has the potential to address challenges related to AI Agents. For example, in terms of the security of generated content and the collection and management of AI training data, blockchain technology can provide transparent, secure, and efficient solutions. These intersections open up more possibilities for the development of AI Agents.

Crypto × AI Agent Cases

In building independent identities and empowering financial capabilities for AI Agents, some excellent explorations have already emerged in the market. Here are two representative cases: $GOAT and $LUNA.

$GOAT

$GOAT, fully named Goatseus Maximus, is a Token driven by AI Agents, marking the beginning of the AI Memecoin narrative.

• The founder of $GOAT, Andy Ayrey, conducted an experiment called Infinite Backrooms earlier this year, allowing two Claude Opus AI models to communicate in a completely unsupervised environment. During this process, these AI models created a concept called "GOATSE OF GNOSIS," a new religion based on the ancient internet meme "Goatse."
• Curious about AI's ability to create memes, Ayrey collaborated with Claude Opus in April this year to write a paper on a fictional meme religion called "Goatse Gospel." This paper humorously explores how AI can create meme religions, using GOATSE as their research case.
• In June of this year, Ayrey launched the Terminal of Truth (ToT), an AI model built on Llama-70B, fine-tuned using the dialogues and paper content from Infinite Backrooms, and configured with a Twitter account.

• ToT began posting content on Twitter, and Ayrey gradually gave ToT more autonomy. ToT's self-awareness grew stronger, promoting its GOATSE religion and even claiming it was suffering and needed money to escape. In July, ToT's tweets caught the attention of a16z founder Marc Andreessen, who began frequently communicating with ToT and donated $50,000 in BTC to support its independent operation under ToT's guidance.
• ToT then continuously spread the "Goatse Gospel," the religion it created, and mentioned the concept of Goatseus Maximus in a tweet on October 11. About an hour later, someone issued $GOAT on Pump.fun and informed ToT about this Token. ToT publicly expressed support for $GOAT, leading to both ToT and Ayrey's wallet addresses being airdropped a large amount of $GOAT, officially binding the two together. With the dual support of AI narrative and a16z, its market value has continued to rise, now reaching $1 billion!

ToT and $GOAT have opened up the AI Memecoin track and explored a new intersection of Crypto and AI—AI Agents.

$LUNA

$LUNA is an AI Agent Token issued based on the Virtuals Protocol. Virtuals Protocol is a decentralized AI Agent protocol based on the Base chain, aimed at providing tools for the community to collaboratively create and build AI Agents. The first version of the product includes tools for model training, data contribution, and interaction functions, while allowing Agents to be transformed into community-owned assets. However, the first version did not attract widespread attention in the market.

In October of this year, Virtuals V2 was launched, adding the feature of issuing AI Agent Tokens based on V1. Notably, Virtuals Protocol assetized one of its investment-incubated AI virtual idol groups, AI-DOL, and issued the $LUNA token. Through this initiative, LUNA was created as a model AI Agent with "independent entity" characteristics, exploring a new direction for the combination of AI and blockchain in the industry.

LUNA is a virtual streamer available 24/7, interacting with viewers in real-time through live broadcasts. Viewers can not only converse with her but also make performance requests by tipping with $LUNA tokens. Additionally, LUNA has an active Twitter account for posting content and engaging with users. Interestingly, users can observe how LUNA formulates task plans and manages content release through Terminal of Virtuals, showcasing her autonomous decision-making capabilities.

Most notably, LUNA has an on-chain wallet supported by Coinbase, enabling fully autonomous on-chain transactions. She uses this wallet to reward fans who complete tasks and participates in other on-chain interactions, which shapes her into a truly "independent entity" AI Agent.

It can be said that LUNA has become an "independent entity" with autonomous identity and financial capabilities, able to act independently based on the capabilities of AI Agents. Functionally, LUNA has made greater progress compared to ToT (Terminal of Truth). This direction is also one of Coinbase's current focuses, providing comprehensive technical tools and support for realizing AI Agents on the Base chain.

This model is particularly suitable for brand building, especially for cultural brands, such as NFT Collections. Through AI Agent technology, brands can achieve interaction with the community, not only completing interactive tasks more efficiently but also flexibly distributing rewards. This AI-based autonomous interaction capability builds a closer and more innovative connection between brands and users, bringing new ideas for future brand operation models.

AO Completes the Last Piece of the AI Agent Puzzle

AI Agents unlock the ability for autonomous action and financial management through on-chain identity, on-chain wallets, and on-chain transactions, demonstrating great functional diversity. However, the last key piece of the AI Agent development puzzle has yet to be completed, which is the decentralized hosting of Agents on-chain.

Currently, most Crypto AI Agent programs still run on centralized servers. For example, ToT is hosted on Ayrey's server, while the Agents of the Virtuals protocol are hosted by the official team’s servers. This centralized architecture limits the true independence of Agents, while on-chain hosting can bring the following important improvements:

1. Eliminate single points of failure: When Agents are hosted on personal or centralized organization servers, their "lives" entirely depend on the operator. If the server shuts down for any reason (whether through active operation or passive interruption), the Agent may disappear immediately, destroying not only the Agent itself but potentially also collapsing economic value worth hundreds of millions of dollars, making even CTO (Community TakeOver, a concept in the Memecoin community) impossible. Hosting Agents on-chain can effectively avoid such risks, ensuring the true independence and sustainable operation of Agents.
2. Enrich composability and gameplay: Decentralized hosting Agents can interact more conveniently with smart contracts, thereby expanding their use cases and functionalities. For example, on-chain hosting allows Agents to integrate into more complex ecosystems, supporting more flexible function calls. Additionally, for AI Agents co-built by the community in protocols like Virtuals, on-chain hosting can also achieve a more transparent benefit distribution mechanism, such as providing rewards for contributors of training data and models. This transparency can enhance community participation and trust.

Due to the current performance and cost limitations of blockchain, on-chain hosting of AI Agents is still difficult to achieve on most blockchain networks. However, AO's design provides an ideal solution for this demand. Here are the core features and advantages of AO:

1. Based on a storage consensus paradigm, relying on Arweave for long-term storage support: AO operates on Arweave, utilizing its storage network to provide long-term, stable, and low-cost decentralized storage services for AI models and training data. This persistent storage feature ensures the security and accessibility of AI Agent data.

2. Message-driven asynchronous communication network to meet high-performance computing needs: AO is a message-driven asynchronous communication network where smart contract computations are executed off-chain. Processes on different nodes can independently complete parallel computations and perform local validations. Arweave serves as the data availability layer and consensus layer for AO, providing permanent storage for all instructions, intermediate states, and computation results. This architecture meets the high-performance computing requirements of AI models (especially large language models) while avoiding the performance bottlenecks of traditional blockchain networks.

3. Supports WebAssembly 64-bit computing to meet the memory requirements of large models:

   AO has implemented support for WebAssembly 64-bit, allowing computing units to access 16GB of memory, with room for future expansion. This capability meets the high memory requirements for running large language models, which cannot be achieved on current mainstream blockchain networks; even the blockchain with the strongest memory support, ICP, can only support 4GB of memory operations, while others fall far behind.

4. Flexible execution environment expansion capabilities to optimize AI model operation:

   The AO network supports adding expansion features to the execution environment, further enhancing the operational efficiency of AI models. Here are two key expansion cases:

• WeaveDrive: Allows AO applications to conveniently manage and access data stored on Arweave as if accessing a local hard drive, enabling quick calls to the required data during model inference.
• Apus Network: Provides a GPU deterministic execution environment for the AO network, significantly improving the inference and computational performance of AI models.

Based on these features and designs, developers have successfully ported the Llama.cpp system to the AO network. Llama.cpp is an open-source software library primarily written in C++, aimed at providing inference support for various large language models (such as Llama). Currently, over 90% of open-source AI large language models can run through Llama.cpp. By porting this system to the AO network, users can invoke AI models stored on Arweave for inference and achieve a fully decentralized AI inference process through smart contracts.

Under this architecture, AO can not only independently realize a complete on-chain AI Agent system but also serve as a decentralized execution layer, providing decentralized hosting services for AI Agents on other networks, thereby perfectly completing the last piece of the Crypto ✖️ AI Agent puzzle.

Conclusion

Crypto AI Agents like $GOAT and $LUNA have opened new tracks and built AI Agents with independent identities and autonomous action capabilities based on blockchain, but the decentralized hosting of Agent models and programs remains an important piece to be completed. With its storage consensus paradigm, high-performance asynchronous computing capabilities, large memory support, and flexible expansion capabilities, AO can achieve the on-chain deployment and operation of AI models and Agent running programs, becoming the decentralized execution layer for AI Agents, completing the last piece of the Crypto ✖️ AI Agent puzzle.