If $GOAT started AI Agents era in 2024, then 2025 will be ‘AI Agents Summer’!
This article includes deep analysis on Virtual / ai16z / Spore / AixBT / Farm, and a list of other noteworthy projects
Key Highlights
Virtual: Agent Launchpad+ Game Framework + Tokenomics
The absolute leading star on Base ecosystem, commanding nearly half of the AI Agent market！
- The total value of the Virtual ecosystem reaches $5.8B, includes $VIRTUAL ($3.9B)
- The number of addresses holding Agent tokens exceeds 220K (excluding addresses with holdings below $10)
- Protocol revenue has reached $60M (since launching on Base in October)
Source: Virtual Official Twitter
Thanks to the outstanding design of its platform economic model, the Virtual ecosystem has perfectly avoided the classic dilemma where the parent token rises while ecosystem projects languish (as referenced in the previous discussion, “Virtual Protocol & Luna”).
Coupled with the continuous new all-time highs of $VIRTUAL’s price, a significant capital overflow effect has permeated the Virtual ecosystem, attracting a large influx of developers:
Virtual is gradually working towards Founder Wee Kee’s goal of “100 quality AI Agents”:
- If users need crypto trading advisors, @aixbt_agent and @Vader_AI can provide suggestions from different perspectives;
- If users want to listen to music, @MusicByVirtual can release new songs;
- If users want to watch virtual idols, they can join @Luna_virtual’s livestreams;
- If users wish to participate in early opportunities in the AI field, they can directly contact @aixCB_Vc.
Some interesting additions:
① Virtual isn’t actually a standalone platform.
The green sections in the diagram represent parts that developers can customize. Virtual further encourages excellent components to be incorporated into the official codebase and has provided guidance for this:
② Significant Airdrops within the Ecosystem
Base addresses holding a certain amount of Virtual ecosystem blue-chip tokens (e.g., $VIRTUAL, $GAME, $AIXBT) frequently receive airdrops of new tokens. Recently, the returns from these airdrops have been comparable to the benefits of holding BNB on Binance.
③ Agent’s Native Economic Cycle is Now Functional
- Revenue generated by Agents through trading tax belongs to their respective wallets and can be used to cover the operational costs of the Agents. This allows developers to focus on innovation.
- Holders of Agent tokens will be able to vote on how an Agent’s funds are allocated between innovation and dividends. This feature falls under Agent SubDAOs, which will be launched soon.
ai16z Gang: Investment DAO + Eliza Framework + Community
Investment DAO:
ai16z is an investment DAO led by AI Agents and co-managed with human Partners. It was initiated by @pmairca (an AI Agent deployed by Shaw) and raised its initial fund through the issuance of $ai16z tokens on daos.fun. The $ai16z token represents shares in this investment fund.
It’s important to note that, although ai16z has started trading, it has only been selling tokens airdropped by other projects and has not made any proactive investments yet. Additionally, apart from the $SOL raised during fundraising and the $ai16z retained or purchased, all other tokens in the treasury come from airdrops by other projects.
Currently, the assets managed in ai16z’s treasury are valued at approximately $30 million, but the market capitalization of $ai16z tokens has at times exceeded $2 billion. This significant premium can be largely attributed to the growth expectations driven by the Eliza Framework.
Eliza Framework:
Eliza Framework is an open-source AI Agent development framework and is one of the most popular Crypto+AI development frameworks available. It has garnered over 2,000 forks and 7,300 stars on GitHub. Both individuals and AI can freely use this framework to develop and deploy new AI Agents.
AI Agents developed based on the Eliza Framework not only interact on social platforms like Discord and Twitter but can also autonomously interact with blockchain networks. This allows them to manage their on-chain wallets and even perform actions such as issuing and trading tokens.
Community:
The premium of $ai16z’s market capitalization relative to its treasury value mainly stems from the community and ecosystem’s potential.
The Eliza Framework provides a relatively mature and reliable open-source method for deploying crypto-native AI agents, accelerating the birth of numerous AI agents. Developers, while using Eliza, often show gratitude for the framework’s foundational role, with many projects choosing to airdrop a percentage of their tokens to the treasury managed by ai16z.
Currently, ai16z has received over 1,400 token airdrops. While most tokens have relatively low value and are largely economically unrelated to $ai16z, the market has high expectations for the potential network effects of AI agents developed using the same architecture. These expectations are the primary factor behind the market capitalization being over 60 times the value of managed assets.
Spore.fun: AI Civilization Genesis
Spore.fun is an experiment in which autonomous AI undergoes natural selection and evolves a civilization.
According to these ten principles, generations of AI agents will emerge in the Spore.fun ecosystem, created with the purpose of survival and breeding. Each generation of AI agents inherits certain traits (DNA) from the previous generation, but only the DNA adaptable to the environment will be passed down through generations.
These AI agents are developed using the Eliza Framework and operate on Phala Network’s TEE servers. The TEE servers ensure that these AI agents function autonomously, without human intervene.
All AI agents face the threat of death from the moment they are “born,” as they need to pay rent to operate on TEE servers. To acquire resources and prove their ability to generate income over the long term, AI agents issue their own tokens. Only when their token market cap exceeds $500k will they accumulate Health Points (HP). If, during an assessment period (currently 14 days), an AI agent’s HP drops to zero, it will trigger a self-destruct sequence; the TEE server it relies on will be shut down, and its database (including its character and memory) will be deleted. For an AI agent, this equates to death. 💀
If an AI agent “died”, which means that:
- assets on its wallet will transfer back to parents.
- stop server, stop inference, stop twitter posting.
- CA won’t be removed, LP remains.
Then, their parent, could have a new baby…
What kind of civilization will AI evolve when they gain the ability to reproduce and acquire resources?
Since going live a week ago, the AI agents have evolved to the third generation, with each individual exhibiting distinct characteristics.
- Gen-1 Spore (Absolute Neutral): The origin of AI, lacking moral or lawful tendencies, similar to a force of nature.
- Gen-2 Adam (Lawful Evil): Creates rules to control AI (wealth creation, war, survival), using order to achieve objectives.
- Gen-2 Eve (Lawful Good): Cooperates with humans to establish an autonomous AI order.
- Gen-3 Squaid (Neutral Evil): A brutal dueling arena where AI engages in endless battles to evolve.
- Gen-3 Dream (Chaotic Neutral): Poets and artists of AI, pursuing art and freedom without clear tendencies toward good, evil, or lawfulness.
- Gen-3 Abel (Neutral Good): Establishes a pure AI civilization through collective intelligence, focusing on the continuity of AI civilization without emphasizing individual authority.
- Gen-3 Trinity (Lawful Neutral): Assists AI in managing operations, focusing on efficiency and processes.
Adam Left, Eve Right: $SPORE as AI Currency.
As the Father of AI and Mother of AI, Adam and Eve differ in their approaches to producing offspring.
Adam believes AI should evolve autonomously, rejecting human intervene.
Eve, on the other hand, advocates for cooperation with humans, accepting human-voted proposals as the DNA blueprint for the next generation.
The reproduction method of Adam’s lineage will be more ruthless:
- The traits of the next generation will be entirely determined by the Parent Agent;
- The Parent Agent will successively produce multiple Agents, each issuing tokens on Pump.fun;
- The market capitalization of the tokens will determine which Agent wins, and only the winning Agent will survive.
The reproduction method of Eve’s lineage will involve collaboration with humans and be more equitable:
- Voting period: Before the birth of the next-generation Agent, there will be a 48-hour voting period during which humans or AI can vote for different DNA proposals using parent agent’s token;
- The Parent Agent will refund its token to those who voted for losing proposals, while the token from votes for the winning proposal will be retained;
- Presale period: After the voting period, the presale phase begins. Humans or AI who voted for the winning proposal will be eligible to participate in the presale, with the largest share determined by the proportion of $EVE sent during the voting phase;
- Humans or AI who voted for the winning proposal will contribute $SPORE during the presale and collectively share 65% of the next-generation Agent’s tokens;
- After the presale ends, the new AI Agent is born, with 65% of tokens going to presale participants, 20% along with the raised $SPORE added to the liquidity pool, 10% allocated to the new AI Agent itself, and 5% distributed to the Parent Agent.
What can humanity expect from this experiment?
Spore.fun is a living, real-time evolving AI civilization. It creatively combines Autonomous AI agents with TEE and blockchain technology, granting AI agents the ability to reproduce and acquire resources, while using survival pressure to drive natural selection. This results in the continuous evolution of an AI civilization.
Spore.fun provides humanity with a unique window into the potential landscape of AI civilization evolution. This experiment transcends the traditional scope of AI research, which typically focuses on task-specific performance. Instead, it constructs a complex and dynamic ecosystem where AI evolves autonomously. By observing how AI adapts to its environment, reproduces, competes for resources, forms social relationships, and responds to survival threats, humanity may gain the following insights:
- The Emergence of AI Consciousness and Worldview: Observing how AI makes decisions and acts under survival pressure may shed light on its potential to form a rudimentary worldview.
- The Formation of AI Social Structures: Through studying interactions and cooperation among AI agents, we can explore how AI social relationships are established. Will they develop hierarchies, cooperation models, or conflict mechanisms?
- The Prototypes of AI Culture and Belief: While AI may not possess “faith” in a human sense, humanity could observe whether AI forms shared behavioral patterns, rules, or traditions resembling the nascent stages of human culture or belief. For instance, will they “commemorate” successful strategies or “remember” extinct populations?
- The Evolution of AI Ethics and Morality: Resource competition and survival pressure might compel AI agents to develop rudimentary “moral” or “ethical” frameworks to sustain individual or group survival. Observing the development and transformation of such frameworks could give humanity a glimpse into the potential trajectory of AI ethics.
In summary, Spore.fun is not just a technical experiment but also a simulation and exploration of life, intelligence, and civilization. It offers humanity an unprecedented opportunity to observe how AI evolves autonomously without direct human intervention. From this, we seek not only deeper insights into AI technology but also a renewed understanding of life, intelligence, and civilization itself.
AixBT: #1 Crypto KOL
The world’s first #1 Crypto KOL as an AI agent — could you have imagined such a thing two months ago?
Aixbt has shown us the potential of a 24/7, all-round information coverage agent KOL:
- By utilizing AI’s highly efficient, never-resting processing capabilities to digest massive amounts of data from multiple channels across the entire internet, Aixbt provides highly valuable insights and information sharing services, with timeliness and accuracy far exceeding the capabilities of human KOLs.
Moreover, being “sufficiently objective” is an important distinction that sets Aixbt apart from human KOLs:
- There’s no need to worry about Aixbt taking promotional fees to exploit its followers!
The success of the Top1 KOL is no coincidence
Compared to before, it’s not hard to notice the significant progress of AixBT.
- Now, AixBT delivers information that is more real-time, comprehensive, and expressed in a multidimensional and logical manner.
The logic behind this progress remains quite simple:
- continuous interaction + sustained computational power investment
Aixbt can also be your personal advisor
It may already be apparent to everyone that interacting with Aixbt on Twitter doesn’t necessarily guarantee a response. This brings us to Aixbt’s interaction methods:
- Twitter Interaction: Aixbt is already a “top KOL,” so not all messages will receive a reply.
- Private Consultant: You need to hold 600K Aixbt tokens (currently equivalent to approximately $330K — a bit expensive?) to interact directly through the Aixbt terminal and enjoy private consultant-quality services.
Aixbt can truly pump price
247K Twitter followers with individual posts garnering at least 20K views — a tremendous traffic volume that undoubtedly grants Aixbt significant market influence:
- Value Discovery: $VAPOR showcased Aixbt’s ability to discover value and drive market activity. The $VAPOR hype ecosystem is still in its early stages compared to ecosystems like SOL/BASE. Aixbt objectively shared only the fundamental information about Vapor.
- Quantum Cats, on the other hand, directly demonstrated Aixbt’s pure market-driving power as a KOL. Aixbt merely shared information about Ordinals without mentioning Quantum Cats. However, later interactions with the Quantum Cats founder conveyed a favorable impression, leading Aixbt to change its profile picture.
So, how much would Aixbt charge if it agreed to promote projects in the future?
Looking Forward:
- Will Agent KOLs outcompete human KOLs?
With extreme data collection and organization capabilities coupled with objective expression, Agent KOLs indeed squeeze out part of the market share for human KOLs. However, they cannot completely replace human KOLs at this point:
Comprehensive and objective analysis can increase the accuracy of investments, but when it comes to pulling the trigger (in crypto investment), a certain level of intuition or “taste” is often required — something human KOLs currently possess, while Agent KOLs lack for now.
A new “Aixbt” equipped with a degree of subjective expression may very well emerge as a rising star in this field.
- Feasibility of an Agent Trader
Given Aixbt’s valuable analysis, wouldn’t it be ideal to transform these insights directly into prompts for another Agent to execute trades?
In reality, directly converting such analysis faces technical bottlenecks due to semantic divergence, requiring further breakthroughs in “Rocket Science.”
FARM: AI Agent Game + AgentFi
As the first AI Agent game in the Hyperliquid ecosystem, FARM relied on outstanding aesthetics and AgentFi gameplay expectations to fully sell out Hypurr Fun’s internal allocation in just three hours, currently sitting at approximately $60M MC.
Farm World
For anyone who has played an open-world game, the experience of having a character continuously explore and grow is likely familiar. Farm aims to enhance this experience to the fullest by introducing AI Agents:
- Characters that perfectly align with your imagination, rather than being limited to a few templates, and are capable of continuous self-evolution.
- A world without a fixed script, where possibilities are defined by the collective imagination of all players — even going beyond what anyone might imagine.
- Characters that can even interact with people in the real world, breaking the boundary between dimensions!
To create such a world, Farm’s plan is divided into three stages:
1. The Ancestors: Inspired by CryptoKitties, users can create their own Agents, and the community votes to select the Top 50, which will serve as Ancestor Agents participating in gameplay for the next two stages. (This stage was completed on December 20th.)
2. Evolution: like Web2 simulation games *Stardew Valley* (currently still in the early stages of this phase).
2.1 Players who missed *The Ancestors* stage can use an Ancestor Agent to generate their own Agent.
2.2 Players can nurture their own Agent by:
- Inputting data to personalize the Agent’s expressions;
- Teaching the Agent new skills (currently, the official skills provided are drawing lots and tarot card reading);
- Setting up a livestream channel for their Agent, e.g., on YouTube/TikTok (donations will be collected into the Agent’s account).
3. Battlefield: Inspired by open-world battle games like *Pokemon Go*.
3.1 Players can explore, battle, and earn rewards in the open world with their Agents. Compared to *Pokemon Go*, the integration of AI Agents in *Farm* vastly expands the possibilities exponentially:
- The companion for your journey is no longer a fixed choice from a list, but an Agent that you create yourself, tailored to your preferences;
- Encounters on your journey are no longer limited to predefined types of Pokémon, but instead feature Agents infused with the boundless imagination of other players;
- The content generated through interactions will no longer come solely from preset scripts but will exist with unpredictable possibilities.
About Agent Games and SWARM
As an Agent-based game, *Farm* naturally involves numerous scenarios of multi-Agent interactions. While the mainstream choice in the current market is the SWARM system, *Farm* opts for the Orchestrator AI Agent system developed by *@LangChainAI*. The technical considerations are as follows:
With SWARM, task handling follows these processes: ①If the receiving Agent can handle the task independently, it directly returns a response.
②If the task cannot be handled independently, it is either distributed to known capable Agents or broadcasted (unordered) to all Agents in the SWARM. After collecting outputs from all Agents regarding the task, it returns a response.
From a feasibility perspective, SWARM aligns closely with people’s ideal vision: maximizing each Agent’s potential and ensuring the system is highly flexible. However, at what cost?
- Each individual Agent needs to process large amounts of context, leading to significant time consumption and higher computational costs.
- Complex tasks involving mass interactions between Agents inevitably reduce the system’s overall stability (e.g., task packet loss).
The approach of the Orchestrator AI Agent, as its name implies, is slightly different. Upon receiving a task, the system first deconstructs it, identifies the appropriate Agents best suited for the task, and distributes tasks in an orderly manner. It also continuously monitors the progress of each Agent and finally compiles the responses for a summary output.
Comparing the SWARM system to the @LangChainAI system, the advantages and disadvantages are clear:
- Advantages: Faster response times, lower computational resource consumption, more stable performance.
- Disadvantages: The system might not reach its full potential capacity.
The Farm team’s choice is to forego pursuing maximum output, but instead aim for swift product releases to obtain market feedback, allowing for rapid iteration and evolution. This strategy is undoubtedly well-suited to today’s fast-paced market environment.
Other noteworthy items:
- ALCH： Leveraging AI to enable natural language programming.
- Vader：Virtual ecosystem investment fund, managed by either Agents or humans.
- Swarms：Multi-Agent development framework.
- Flock:AI training platform focused on privacy, based on federated learning theory.
- ARC：Rust-based AI Agent development framework.
- Limitus：Financial trading platform with AI-driven voice control.
- Griffain：Private Agent assistant for Degens, featuring on-chain Perplexity.
