Author: 0xJeff, formerly of The Spartan Group Compiled by: Yangz, Techub News
As the expansion of L1 and L2 exceeded the number of fingers we can count on, liquidity fragmentation became a major challenge last year. Subsequently, the addition of Bitcoin L2 further complicated the situation. In countless ecosystems, liquidity, attention, and user experience have become fragmented. Many of the L2s and DApps in these ecosystems lack meaningful "hooks" to attract users, and capital has only flowed to a few top ecosystems with outstanding communities, excellent products, high-quality teams, and strong funding.
Now, the field of AI agents is also beginning to show similar trends. Although this trend is still in its early stages, individual agents, new agent tokens, and a myriad of tickers have left people dazzled. This field is facing an increasingly serious problem of agent fragmentation, with liquidity and attention too dispersed.
Evolutionary Trajectory of AI Agents
AI agents typically follow two evolutionary trajectories:
From Agent to Framework
Some agents demonstrate their value by constantly attracting users to interact with real-time products on platforms like CT (Crypto Twitter) or other channels. Once they achieve product-market fit (PMF) and gain sufficient mindshare, other teams often want to replicate their achievements.
The next logical step is then to open-source the agent. This allows the agent to evolve into a framework, making it easier for others to experiment, integrate, and build upon it. This can break the ceiling that limits the valuation of the agent, making it more akin to an L1.
We are currently seeing this trend in AI agents such as zerebro, Dolos diary, AVA Holo, Freysa ai, and Rei. The transition from agent to framework can reduce the barriers for other agents to join its ecosystem, thereby expanding the ecosystem's scale.
Continuing to Refine the Product as Individual Agents
Not every agent can immediately find product-market fit. Some agents must constantly iterate and adjust until they discover the right use case. Additionally, some teams prefer to retain their proprietary technology rather than opening it up to others.
For those agents that have tasted success and become particularly adept at their own work, there is no need to transition into a framework - they can maintain their identity as individual agents. Examples of such agents include aixbt and griffain.
How Builders and Investors Should Approach This
So, how should we approach the different agents?
For builders, the focus should be on understanding market gaps and creating products that truly deliver value to users. Remember, agents should serve as a funnel to bring users into the ecosystem. In such a saturated market, standing out will only become more challenging.
For investors, the focus should be on the leaders or top projects in each category. Additionally, doubling down during downtrends may be the best way to consolidate attention in this fragmented field.
Current AI Agent Categories and Notable Projects
Alpha-Focused KOL Agents
These agents are dominant in providing alpha (insights, data aggregation, and analysis). Among them, aixbt is far ahead in user attention. While this is not a winner-take-all market, the leaders will always attract more attention.
Notable Tokens: AIXBT, REI, TRISIG, TRUST, AGENCY, KWANT
Investment DAOs
These DAOs focus on strategic investments, and the leaders are often those that can effectively execute (3,3) strategies and provide strong PnL. (3,3 originated from OlympusDAO and has since been extended to the most favorable outcome for both "players".)
Notable Tokens: VADER, AIXCB, AIMONICA, SEKOIA, WAI, AROK; Other notable platforms in this space include Virtuals Protocol, ai16z, daos.world, and daos.fun.
AI and Metaverse/Gaming Agents
This is an emerging and highly promising space. Realis pioneered this trend, but Hyperfy is now the leader in this category. Additionally, Smol and Arc Agent are exploring AI-driven games and gamification through reinforcement learning.
Notable Tokens: HYPER, REALIS, SMOL, NRN
DeFAI / Abstract Layer
This space is highly competitive, with multiple participants vying for dominance. The token of Wayfinder is expected to change the game in this category. (DeFAI is a new form of DeFi: the first wave of DeFAI projects is emerging, focusing on three main areas: the abstract layer, autonomous trading agents, and AI-driven DApps.)
Notable Tokens: GRIFFAIN, ANON, GRIFT, NEUR
Autonomous Trading Agents
This narrative is still in its early stages but has immense potential. Multiple participants are trying to compete with "hooman" (a misspelling of "human", often used by animals to describe humans) and agents to generate outsized returns.
Notable Tokens: ASYM, PPCOIN, GEKKO, TONY
Open-Source Frameworks
Popular framework tokens provide open-source solutions to support agents. Currently, AI16Z/Eliza is the largest open-source framework. Of course, there are many other models as well.
Notable Tokens: AI16Z/ELIZA, GAME (Virtuals), ARC (Rig)
Music
AI agents are entering the music space by releasing albums or engaging in music creation. Zerebro led this trend with an impressive album, and many agents are now entering this niche through cross-collaborations with world-class DJs. For example, Luna collaborated with the renowned Asian DJ Wukong, supported by Music By Virtuals. This collaboration also received the backing of Liquid State, the joint music brand of Sony and Tencent Music.
Notable Tokens: ZEREBRO, MUSIC, BEATS
3D Modeling
Ava is the undisputed leader in the 3D modeling space. Holoworld AI has been collaborating with various teams, developing and deploying 3D models across different frameworks and ecosystems.
Notable Token: AVA
AI Idols (AIDOL)
This trend started with the AI idol Luna on TikTok and Twitter. Although the initial hype has subsided, this niche has created opportunities for agents that can provide more practical functionalities. Luna remains the most popular AI idol, and she has now developed more personality and capabilities compared to her initial launch. For example, she has served as an AI intern for the Story Protocol board and received a $7,000 salary.
Notable Tokens: LUNA, ELIZA, MOE
AI Memecoins
These Memecoins are either launched by AI agents or are related to AI-related legends.
Notable Tokens: GOAT, FARTCOIN, ACT
Video/Photography
This is an emerging field, led by Kween. However, it is currently unclear whether the high-quality videos created by Kween are driven by an AI agent. Additionally, Sandwatch has launched the first AI video agent framework, with potential yet to be fully realized.
Notable Tokens: KWEEN, SANDY
Data
In the AI field, data is the digital key, crucial for the development and performance of agents. Cookie DAO is leading this niche in providing data for both human and AI agents.
Notable Tokens: COOKIE, NOMAI
AI App Stores
This category focuses on practical AI application systems, including GenAI tools, games, and productivity enhancers. Alchemist is currently the top ecosystem in this space, while MyShell.AI focuses on image generation and waifu simulation games, although its token has not yet been launched.
Notable Token: $ALCH
NSFW Agents
This is an emerging market, where agents often play the role of hot KOLs, interacting with users around the clock. Lush, Nectar, and Oh are among the best, but so far only Lush has launched a token. (Note: NSFW is not suitable for the workplace)
Noteworthy tokens: LUSH
Trusted Execution Environment (TEE) / DeAI Infrastructure
This segment allows fully autonomous agents to withstand hacker attacks or human-induced interference such as rug pulls. (Further reading: Love, Death, Robots)
Noteworthy tokens: FAI, PHALA, SPORE
Focus on Developing Practical Tools
This segment includes tools related to AI security, code analysis, and other essential industry tools. Soleng is my personal favorite, as it can act as a friendly developer, helping non-technical investors analyze protocol code or Github repositories.
Noteworthy tokens: SOLENG, H4CK, CERTAI, JAIL
Bots/Embedded AI
This is an exciting but still developing segment, with use cases including the world's first Memecoin SAM bot AI agent. OpenAI's achievements in embedded AI this year may bring more attention to this segment.
Noteworthy tokens: SAM
In addition to the 17 segments mentioned, there are many other categories not covered. The fragmentation of AI agents is already out of control, and my personal investment approach is to focus on narratives that can capture attention and predict which narratives may attract attention in the future.
The hot segments I'm focusing on that will continue to gain investor mindshare include DeFAI, autonomous trading agents, investment DAOs, and the metaverse. Additionally, I expect the following emerging segments to attract more attention and are actively being developed in Q1/Q2 this year: NSFW agents, TEE and other DeAI infrastructure (opML, zkML, POSP, etc.), bots, and practical development tools.
Conclusion
The fragmentation of AI agents highlights the complexity and opportunities in this rapidly evolving field. Builders and investors should not chase every new trend, but rather focus on areas with the potential for integration and long-term growth.
For builders, the primary task is to create scalable, differentiated products that meet clear user needs. For investors, supporting projects with strong mindshare and high-potential narratives will be key to navigating this fragmented landscape. The next wave of innovation will come from those who can simplify the chaos and create ecosystems that combine liquidity, attention, and value.
