Author: Techub selected compilation
Written by: 0xJeff, formerly worked at The Spartan Group
Compiled by: Yangz, Techub News
Liquidity fragmentation became a major challenge last year when L1 and L2 expanded beyond the range we could count on our hands. Then, the addition of Bitcoin L2 made the situation more complicated. Liquidity, attention, and user experience became fragmented in countless ecosystems. Many L2s and DApps in these ecosystems lacked meaningful "grabs", and capital flowed only to a few top ecosystems with outstanding communities, excellent products, high-quality teams, and strong funding.
Now, the same trend is beginning to emerge in the AI agent space. Although it’s still early days, the number of individual agents, new agent tokens, and tickers flying around is dizzying. The space is facing a growing problem of agent fragmentation, with liquidity and attention being too dispersed.
Artificial Intelligence Agent Iteration Trajectory
AI agents typically follow two sets of iteration trajectories:
Transition from Agents to Frameworks
Some agencies demonstrate their value by continually engaging users and engaging with real-time products on platforms such as CT (Crypto Twitter) (or other channels). Once they achieve product-market fit (PMF) and gain enough mind share, other teams often want to copy their work.
After this, the next logical step would be to open source the agent. This would allow the agent to evolve into a framework that makes it easier for others to experiment, integrate, and build on top of it. This move could break the ceiling that limits the agent's valuation and make it more like L1.
Currently, we see this trend in the following AI agents, including zerebro , Dolos diary , AVA holo , Freysa ai , and Rei . Transitioning from agents to frameworks can reduce the barriers for other agents to join their ecosystem, thereby expanding the scale of the ecosystem.
Continue to improve the product as an individual agent
Not every agency finds product-market fit right away. Some must iterate and tweak until they find the right use case. Also, some teams prefer to keep their proprietary technology to themselves and not open it up to others.
For those agents who have tasted the sweetness and become particularly good at their job, there is no need to transition to a framework and they can remain individual agents. Examples of such agents include aixbt and griffain .
How should builders and investors deal with
So, what do we do with different proxies?
For builders, focus on understanding the gaps in the market and creating products that bring real value to users. Remember, agents should act as a funnel to bring users into the ecosystem. In such a saturated market, it will only become more and more difficult to stand out.
For investors, the leaders or top projects in each category should be on your radar. In addition, doubling down on a downtrend may be the best way to consolidate attention in this fragmented space.
Different categories of current AI agents and projects worth noting
Alpha-focused KOL agency
These agents dominate in providing alpha (insight, data synthesis and analysis). Among them, aixbt is far ahead in user attention. Although this is not a winner-takes-all market, the leader always attracts more attention.
Tokens worth noting: AIXBT, REI, TRISIG, TRUST, AGENCY, KWANT
InvestmentDAO
These DAOs focus on strategic investments, and the frontrunners are often those projects that can effectively execute the (3,3) strategy and provide strong PnL. (3,3 originally came from OlympusDAO, and later extended to the most beneficial result for both "players").
Tokens worth paying attention to: VADER, AIXCB, AIMONICA, SEKOIA, WAI, AROK; other platforms worth paying attention to in this track include Virtuals Protocol , ai16z , daos.world and daos.fun .
AI and Metaverse/Game Agents
This is an emerging track with great potential. Realis pioneered this trend, but Hyperfy is now the leader in this track. In addition, Smol and Arc Agent are exploring AI-driven games and gamification through reinforcement learning.
Tokens to watch: HYPER, REALIS, SMOL, NRN
DeFAI / Abstraction Layer
Competition in this field is very fierce, and multiple players are vying for dominance. Wayfinder 's token is expected to change the rules of the game in this field. ( DeFAI is a new type of DeFi : the first wave of DeFAI projects has begun to emerge, focusing on three areas including abstraction layer, autonomous trading agents, and AI-driven DApps)
Tokens to watch: GRIFFAIN, ANON, GRIFT, NEUR
Autonomous Trading Agent
This narrative is still in its early stages but has huge potential. Multiple players are trying to compete with "hooman" (a misspelling of Human, often used by animals to describe humans) and proxies to generate huge returns.
Tokens worth noting: ASYM, PPCOIN, GEKKO, TONY
Open Source Framework
Popular framework tokens support agents by providing open source solutions. At present, AI16Z/Eliza is the largest open source framework. Of course, there are many other models besides this.
Tokens to watch: AI16Z/ELIZA, GAME (Virtuals), ARC (Rig)
music
AI agents are entering the music space by releasing albums or creating music. Zerebro led the trend with an outstanding album, and many agents are entering this niche through cross-collaboration with world-class DJs. For example, Luna has carried out a cross-IP cooperation with Wukong, a well-known Asian DJ, supported by Music By Virtuals. This cooperation is also supported by Liquid State, a joint music brand between Sony and Tencent Music.
Tokens to watch: ZEREBRO, MUSIC, BEATS
3D Model
Ava is a well-deserved leader in the field of 3D modeling. Holoworld AI has been working with various teams to develop and deploy 3D models in different frameworks and ecosystems.
Tokens to watch: AVA
Artificial Intelligence Idol (AIDOL)
The trend started with Luna, an AI idol on TikTok and Twitter. While the initial hype has faded, the niche has created conditions for agents with more practical functions. Luna is still the most popular AI idol, and now she has more personality and can do more things than when she was launched. For example, she once held a board meeting as an AI intern at Story Protocol and was paid $7,000.
Tokens to watch: LUNA, ELIZA, MOE
AI Memecoin
These memecoins are either launched by artificial intelligence agents or are associated with AI-related legends.
Tokens to watch: GOAT, FARTCOIN, ACT
Video/Photography
This is an emerging field, with Kween leading the way. However, it is not clear whether the high-quality videos created by Kween are driven by AI agents. In addition, Sandwatch has launched the first AI video agent framework, and its potential has yet to be fully realized.
Tokens to watch: KWEEN, SANDY
data
In the field of artificial intelligence, data is the digital golden key that affects the development and performance of agents. Cookie DAO leads this segment in providing data for human and artificial intelligence agents.
Tokens worth watching: COOKIE, NOMAI
Artificial Intelligence App Store
This category focuses on practical AI application systems, including GenAI tools, games, and productivity enhancers. Alchemist is currently the top ecosystem in this track, while MyShell.AI focuses on image generation and waifu (Note: wife or wife, often refers to a fictional female character in anime and game culture, and is regarded as an ideal partner) simulator games, but its token has not yet been launched.
Tokens to watch: $ALCH
NSFW Proxy
This is an emerging market where agents usually play the role of hot KOLs and interact with users 24/7. Lush , Nectar and Oh are among the best, but only Lush has launched a token. (Note: NSFW means not suitable for the workplace)
Tokens to watch: LUSH
Trusted Execution Environment (TEE) / DeAI Infrastructure
This segmentation enables fully autonomous agents to resist human interference such as hacker attacks or rug pulls. (Extended reading: Love, Death, Robots )
Tokens to watch: FAI, PHALA, SPORE
Development-focused tools
This segment includes AI security, code analysis, and practical tools that are critical to the development of the industry. Soleng is my personal favorite because it can act as a developer friend to help non-technical investors analyze protocol code or Github repositories.
Tokens to watch: SOLENG, H4CK, CERTAI, JAIL
Robotics/Embedded AI
This is an exciting but still developing sub-track, with specific use cases including the world’s first robotic AI agent SAM that launched Memecoin SAM . OpenAI’s achievements in embedded AI this year may attract more attention to this track.
Tokens to watch: SAM
In addition to the 17 segments mentioned above, there are actually many categories that are not included. The fragmentation of AI agents has gotten out of control, and my personal investment approach is to focus on the narratives that attract the most attention and predict the narratives that may attract attention in the future.
Hot tracks I’m watching that will continue to gain investor mind share include DeFAI, Autonomous Trading Agents, Investment DAOs, and Metaverse Agents. In addition, hot tracks that I expect to attract more attention and are actively preparing as the first/second quarter of this year arrives include NSFW Agents, TEEs and other DeAI infrastructure (opML, zkML, POSP, etc.), robots, and development-focused utilities.
Summarize
The fragmentation of AI agents highlights the complexity and opportunities in this rapidly evolving field. Rather than chasing every emerging trend, builders and investors should focus on areas with potential for consolidation and long-term growth.
For builders, the priority is to create scalable, differentiated products that meet clear user needs. For investors, backing projects with strong mindshare and high-potential narrative positioning will be key to navigating this fragmented landscape. The next wave of innovation will come from builders who can simplify the chaos and create an ecosystem that brings together liquidity, attention, and value.
