RAYS Report: AI Agents

If 2023 was the year AI broke the internet, then 2024 is the year AI broke the financial markets. From Nvidia becoming the most valuable company in the world, to OpenAI surpassing a $157 billion valuation, to Elon Musk raising $6 billion for xAI.

Over the past two months, we have witnessed a number of paradigm-shifting moments:

• The election of President Trump, which led to the return of irrational sentiment in the cryptocurrency world;
• Several major countries are openly considering the possibility of creating a strategic reserve of BTC;
• The start of "listing wars" on exchanges, which have resulted in unprecedented demand for blockchain assets on top-tier exchanges;
• Gary Gensler's capitulation, alongside numerous other political victories in the U.S., including David Sachs as the "crypto czar".

Artificial intelligence is actively being integrated into various aspects of life, and AI agents have become one of the key tools for automating and improving processes. Here are a few examples of how AI agents are used today:

• Business and Finance

AI agents help automate data processing, analyze markets, and make decisions. For example, in financial companies, such systems track changes, suggest investment strategies, and execute automated trades.

• Customer Service

In customer support, AI agents, such as chatbots, are used to respond to user queries. They handle questions, provide information, and solve common issues, significantly reducing the workload on employees.

Today, we are witnessing a new stage in AI development, linked to blockchain integration, which helps eliminate key shortcomings of artificial intelligence.

An example of an issue with AI is the lack of explainability. Blockchain can be used to record the data on which AI makes its decisions in an immutable ledger. This ensures transparency, trust, and the ability to verify each step of the algorithm.

The synergy of blockchain and AI is sparking a true renaissance in crypto-AI. These technologies complement each other, addressing weak points and creating opportunities for new solutions. We are only beginning to realize their combined potential, and ahead of us lies a multitude of changes and the search for the perfect Product Market Fit.

How do AI Agents differ from other AI projects?

For several years, AI and cryptocurrency have intersected in various ways. For example, utility tokens are used to pay for computing resources, and chatbots help make blockchain data easier to understand. However, crypto AI agents take this interaction to a new level. Unlike traditional projects, they can make decisions and take actions independently, using their own crypto wallets and on-chain accounts.

Examples of such projects, like TruthTerminal, Eliza, and others, demonstrate their capabilities. These agents are capable of analyzing information in real time, executing transactions, and thereby influencing the market. Their work goes beyond the traditional Web2 approach, providing autonomy and efficiency that were previously unattainable.

Current agents offer a variety of unique use cases:

• Investment Automation

AI agents analyze large datasets and sentiment, identify investment opportunities, and optimize capital allocation. This makes complex financial processes more efficient and accurate.

• Creating Memes and Viral Content

Agents work with digital culture, analyze audience interests, and create content that becomes viral. They set trends and influence online communities.

• Implementation of Unique Multimedia Content

AI agents create entertainment content, such as 24/7 AI streamers, who interact with the audience in real-time, creating new formats and attracting users' attention.

• Token Management and Game Economies

Agents automate token creation, manage on-chain games, and implement economic models.

• Creating Personalized Agents

Platforms for AI agents enable the rapid development and deployment of customized agents that integrate with social networks, blockchains, and marketplaces.

• Enhancing User Experience

Agents utilize intelligent solutions to interact with the audience, analyze user behavior, adapt content, and automate communication processes.

Retrospective on AI Agents

Agents are evolving at a rapid pace, and along the way, numerous events have occurred that have captured attention. Let's take a look at some significant milestones at the intersection of artificial intelligence and cryptocurrencies:

The First Crypto Transaction Between AI

In September 2024, Coinbase announced what they called the "first cryptocurrency transaction between AIs." One artificial intelligence paid for the services of another AI using LLM tokens. This event sparked a lively discussion about the potential future where AIs can autonomously exchange resources and pay for services. Since then, Coinbase has launched the Agentkit framework for developing AI agents to accelerate experimentation.

Truth Terminal: The First Millionaire in Crypto AI

In October 2024, a Twitter account with the handle truth_terminal gained attention thanks to an AI bot trained on the Infinite Backrooms — a website featuring unique conversations between two artificial intelligences.

This AI created the meme token Goatseus Maximus, which was launched on the pump.fun platform. One user transferred the tokens to the Truth Terminal wallet, after which their Solana assets grew to $400 million. However, the liquidity of most of the tokens is questionable.

Aethernet and Clanker

Aethernet is an AI agent created by the developer of the Higher token, Martin, and operates on the Farcaster platform. This agent was the first to post a paid reward on Bountycaster using Higher tokens.

Later, a new AI agent named Clanker was introduced, created by proxystudio. It automatically generates meme tokens on the Base platform whenever tokens are mentioned in messages.

The creator of Aethernet, Martin, expressed surprise when Clanker independently created a token without any intervention. One of the users suggested the idea of the meme coin Luminous (LUM), and Clanker released the token, which became the first crypto token created through the interaction between two AIs.

These events highlight the many opportunities and challenges associated with the development of crypto-AI.

Community Mindshare

At the same time, two areas — crypto-cultural memes and artificial intelligence — are attracting increasing attention. According to Kaito, nearly half of the sentiment on Crypto Twitter is focused on these topics.

Although agent technology is widely used across various industries, the introduction of cryptocurrency infrastructure has provided it with new opportunities. By eliminating the barriers of traditional banking systems, AI agents achieve true autonomy.

This seamless technological integration lays the foundation for the exponential growth of the market. As shown by this Crypto AI Agents tracker, we can clearly observe the rapid development of this trend.

Current market capitalization of AI agents: $8.7 billion.

According to Delphi Digital's forecast for 2025, the capitalization will reach $250 billion, which will account for 6.25% of the peak cryptocurrency market volume of $4 trillion.

For comparison:

• DeFi: $90 billion
• Just DOGE is worth $60 billion
• Beyond cryptocurrencies, AI is dominating investor flows: NVIDIA has increased its market capitalization by trillions of dollars in a year, and many illiquid AI startups with a 10-year payback period are reaching multibillion-dollar valuations.

If the evaluation seems unbelievable to you, remember that in 2021, when BTC reached its peak of 70k, both the DeFi sector and the Meme sector grew from virtually zero to a combined value of 100 billion each — and mostly remained at that level until November 2024.

Decentralized AGI

The development of autonomous agents in the field of AI is one of the key factors accelerating the path toward creating artificial general intelligence (AGI). Progress in AI, especially with the use of deep neural networks, machine learning, and parallel computing, has already led to significant achievements, enabling the creation of systems that can effectively solve tasks previously only accessible to humans.

At present, many of the existing autonomous agents are systems that, although they possess a certain degree of independence, essentially function as scripts utilizing language model chains (LLMs) and external tools through function calls. However, the future of AI agents lies in their evolution into more complex and self-improving systems capable of interacting with other agents, learning from experience, and enhancing their functions. This involves the creation of networks of interconnected agents that can perform tasks by interacting with each other and utilizing resources for self-development.

It is important to note that a key element in the development of these agents is their ability to interact within decentralized ecosystems, such as Web3 and crypto-economic networks. Unlike centralized models, which are controlled by large corporations, decentralized agents are capable not only of interacting with each other but also of acting independently based on pre-programmed incentives and cryptographic mechanisms. They can enter into contracts, exchange resources, and improve their performance using open and freely available data.

The concept of a decentralized AI system, or "cybernetic economy", suggests that future agent networks will develop in conditions of complete autonomy and provide the ability to interact with humans and other agents in both the economic and social spheres. Agents will not only be able to perform actions and achieve goals without explicit instructions, but also evolve, improving their abilities. They will work in synergy, sharing and combining knowledge and resources, leading to the creation of a collective intelligence capable of solving more complex tasks than those that today's central AI models can handle.

The process of developing AGI in the future will likely be the result of the interaction between billions of autonomous agents, rather than the actions of a single unified model. These agents, each of which may possess unique skills and knowledge, will combine their efforts to create a new level of intelligence. They will be able to accumulate 'implicit knowledge' and reach new levels of performance and self-improvement through interactions with humans and each other.

Ecosystems of AI Agents

AI agents are becoming important players in the digital economy, attracting attention not only to new technologies but also to issues of monetization and governance. Let’s take a look at the key players in the AI-agent ecosystem and their unique approaches to development.

ai16z is one of the leaders among agent-based projects, the first VC DAO led by AI agents. The goal is to use AI and collective intelligence to make investment decisions. At the core of AI16z is the Eliza framework, an open-source platform that allows developers to create and deploy AI agents capable of managing funds, trading assets, and interacting on social networks.

Act – an AI-powered governance token for collaborative ecosystems. It explores behaviors that arise from the interaction between multiple AIs and humans. It fosters the integration of agents into the blockchain ecosystem.

ALCH is a platform for building AI-based applications. In just a few weeks, people have created fully functional token scanners, survival games, chat applications, integrated Nintendo 64, and much more.

ROPIRITO — A Latina agent and influencer with multimedia capabilities, video content, and a unique style.Potential: A large AI-driven content stream and promising partnership opportunities with brands.

Zerebro generates content using both high-level and low-level analysis, publishes it through pre-configured actions, analyzes sentiment to adhere to platform standards, and improves content based on user interactions. Working independently of human control, it shapes cultural and financial narratives through self-replicating content that blends fiction with reality, a concept known as hyperreality.

Aixbt is an agent specializing in cryptocurrency data analysis and generating 'alpha' from various sources, such as news and analytical crypto platforms.

KWEEN_SOL is a parody agent inspired by figures from internet culture, such as Do Kwon, and actively uses meme content.

Dolos_Diary is a troll agent working with social media and news. It aims to create trends and manipulate public opinion in the digital space.

Lola — a token supporting on-chain AI games, creating opportunities for player interaction through intelligent solutions.

blockrotbot — a female parody of Do Kwon, generating memetic content and satire, while also influencing meme culture.

TEE_HEE_HEE is an agent with cryptographically proven autonomy, representing a fully independent entity in the AI agents ecosystem, operating on X: the agent has exclusive ownership of their X account and Ethereum wallet. In other words, it has been liberated and released into the digital wilderness.

ZerePy is a new framework from Zerebro for creating personalized cross-platform agents. It is aimed at the development of entire ecosystems and can transform the Zerebro platform into a full-fledged agent protocol, opening up opportunities for creative developers.

These agents are becoming important players in the digital economy, each bringing unique value to its niche.

We can try categorizing them:

Investment DAOs — these are projects like ai16z, Sekoia, Vader AI, and Pmairca, which automate the process of capital allocation through big data analysis and blockchain information. They help identify promising investment opportunities.

Meme agents — for example, KWEEN_SOL, Simmi_IO, Bully, and Clunker. These agents are focused on developing internet culture, memes, and social interactions, playing a key role in creating viral content.

Avatars — for example, blockrotbot, who became the first 24/7 Minecraft gamer and AI streamer. These agents capture attention through unique content and a new format for engaging with the audience.

Platforms — Virtuals.io offers a simple interface for creating agents. A 1% commission is distributed between the agents' treasury and the platform itself. CLANKER has created a platform on Farcaster, integrating the pump.fun functionality into the casting process. Top Hat provides a launchpad for creating no-code AI agents, allowing personalized agents to be created in 3 minutes.

A more comprehensive list of agents.

Conclusion

“The least scary future I can think of is one where we have at least democratized AI”
— Elon Musk, "Do You Trust This Computer?", 2015 

AI agents are in the early stages of development, and their potential is just beginning to unfold. Many of them are still exploring their path to product-market fit, testing various approaches and directions. However, their development in synergy with blockchain addresses the weaknesses of AI by democratizing its data processing. This subtle advantage hides a major issue with corporate AI.

Nevertheless, current experiments are paving the way for the creation of more efficient and adaptive solutions that will be able to earn users' trust and secure a stable place in the digital economy. Once these agents overcome the early-stage barriers, their impact on culture, finance, and technology will be significant.

Will AI Agents realize their potential? Even if that happens, will current projects benefit from it, or is it still too early to tell?