Author: Edgy Source: X, @thedefiedge Translation: Shan Ouba, Golden Finance

Currently, There may not be a full-scale altcoin season, where all altcoins rise together. The altcoin market is fragmented – there are too many tokens on the market right now and not enough liquidity to support all token gains (even with a full return of retail investors).

Only certain tracks will have full copycat seasons. Now, the AI ​​agent track is at this stage. While the market is generally falling, this sector is hitting all-time highs. AI agents are attracting market liquidity and attention.

If you missed the DeFi Summer in 2020, or the Alt Layer 1 market in 2021... the market is giving you another opportunity! Don't try to buck the trend, don't be clever.

If you've been watching from the sidelines lately, maybe you're wondering if it's "too late" "Admission? I don’t think so:

•The current market is generally depressed, but AI infrastructure and agent projects are reaching new all-time highs. Just imagine what will happen when the market rebounds?

•The entire AI agent track has only been launched for about 2 months (the starting point was the launch of GOAT). Currently these AI agents are in their "most elementary" state. Over time, we're unlocking new features every day.

Imagine how much change it will bring when AI agents can optimize your portfolio and trade directly on the chain?

• Think of AI infrastructure as Alt Layer 1 and AI agents as DApps. Some AI infrastructure protocols may reach Alt Layer 1 valuation levels ($10 billion+).

The main difference is: most AI infrastructure projects do not have VC support, so you don’t need to worry about selling pressure caused by large-scale token unlocking.

•Currently, the top AI protocols have not yet been launched on CEX (centralized exchange). I have a few friends around meFriends are interested in this track, but have encountered some difficulties in entering.

I will mainly discuss AI infrastructure projects because they currently have the best risk/reward ratio.

Instead of picking a single AI agent, it is better to directly invest in the entire infrastructure, which is equivalent to "owning the entire casino." During the gold rush, the richest people were those who sold shovels and picks.

Let’s dive into this track.

Two kings of AI infrastructure

1. Virtuals (@virtuals_io) - the current king

•Market value: $4.6 billion

•Position: Leader in the field of AI agents Main Launchpad platforms

Virtuals has become the preferred launchpad for AI agent projects and currently dominates the Base ecosystem, but may expand to other chains in the future. Its co-founder @ethermage hinted that it is already preparing to launch the platform on Solana, with plans to expand to @HyperliquidX and @AbstractChain in 2025.

Virtuals recently announced a series of strategic goals, such as attracting top AI agent developers and improving technology to build Virtuals into the fairest stage for agent investors. Will Virtuals launch a revenue sharing mechanism for AI agents?

Three core advantages:

1. Token Economic Model: Creating an agent requires $VIRTUALS, and the liquidity matching of all new agents also requires $VIRTUALS. Almost all new proxy launchpads are imitating their token economic model.

2. Network effect: Virtuals currently attracts great attention from the market.Launching agents on the platform means being able to ride on the coattails of the Base ecosystem.

3. Star agent: @AIXBT is the first agent with a market value exceeding US$500 million.

While I don't like making price predictions, Virtuals may break into the top ten by market cap this cycle.

2. Ai16z ($AI16Z) - The King of Challengers

< p style="text-align: left;">•Market value: $2.2 billion

•Status: The birthplace of the ElizaOS proxy framework and the most popular on Github One of the code bases.

Ai16z has experienced many twists and turns since its establishment, but its technical and community influence continues to grow.

The biggest problem was that its token economic model failed to effectively capture value. Now, they are undergoing a massive overhaul and will launch a new Launchpad in the first quarter of 2025.

Starting a project on the Eliza framework will require:

• Pay a startup fee (part of which will Flow back to $AI16Z through buyback and destruction)

•Hold $AI16Z to gain access and liquidity matching opportunities

•Project tokens need to be aligned with $AI16Z Liquidity pairing to stabilize the market

It's like owning a stake in a casino, with each new agent paying an "entry fee".

Strategy and plan:

• Diversified treasury strategy: diversify investments into major L1 generations coins, stablecoins and project tokens, and provide liquidity for the $AI16Z trading pair to ensure long-term sustainable development and avoid sell-off risks.

•Agent Swarms: AI agent clusters will be the focus of the future. Multiple agents work together. This is what decentralized general artificial intelligence (AGI) ) comes true.

•Ecological incentives: Developers building on Eliza will receive:

•Ecological incentives Fund support

•Fee sharing opportunity

Build $AI16Z into the entire decentralized AI The “glue” of the stack, similar to what $ETH is to the Ethereum ecosystem, or $SOL for Solana.

Who is stronger, Virtuals or Ai16z?

Virtuals advantages:

• Strategically forward-looking, at the forefront of the industry, competitors are still catching up.

•The support of Base ecology provides strong momentum for its growth.

Ai16z advantages:

•ElizaOS framework technology is leading, and the experience of developing agents is better.

•Token economic model reform will solve the problem of value capture.

The crypto industry has always liked to take sides, but you don’t have to. I'm not sure who will ultimately win this race, so it might be the easier option to simply hold both. However, I am more familiar with Virtuals and its ecological configuration.

Many industries will eventually evolve into a battle between two heroes, such as Apple and Android, Coca-Cola and Pepsi-Cola. Will this track also face a similar situation? Let's wait and see.

Competitors

Virtuals and Ai16z are dominating the market, but the game has only just begun. If you're looking for higher returns, there are some more "flexible" options to consider.

ZereBro (market cap $632 million) - Transformation Already

ZereBro started out as a general-purpose AI agent that could also create music. But now it is transforming into a complete full-stack framework for creating other agents.

ZerePy is an open source proxy framework based on Python, and the status of Python is self-evident:

1.Python is the most popular programming language in the world.

2. The vast majority of AI tools and libraries are written in Python.

You can use it to start a proxy like ZereBro.

The Zentients agent platform will be launched in early 2025. This new Launchpad will directly empower $ZEREBRO, and each agent created will have value flowing back through launch fees, Liquidity Pairing (LP) bonding, and an innovative liquidity locking mechanism.

ZereBro has now joined Virtuals’ thriving ecosystem and expanded its reach by launching the ZereBro/Virtual liquidity pool on @Base.

They are following the successful strategies of current industry leaders, but their current valuations are much lower than their competitors.

$ARC (market cap $358 million) - Rust-based AI agent framework

Arc.fun An open source agent framework called Rig is being built, which is currently the only agent framework built on Rust. Rust is Solana’s native language, which is key.

Why is $ARC worthy of attention?

•The Rust language optimizes performance through zero-cost abstraction.

•Rig framework in memory and cold startIt performs better than Langchain in terms of running time.

•Core contributor @0xCygaar has started to contribute code to Rig Github. He is also one of the core developers of Abstract Chain.

In short: Arc = Rust + great code base.

$GRIFFAIN (market cap $457 million) - AI app store on Solana

Griffain allows you Agents can autonomously complete on-chain transactions by searching and performing any operation on the chain through natural language.

Core highlights:

•Griffain agent can automatically execute preset transactions on DEX, Such as hourly fixed investment (DCA) or limit order (Limit Order).

•Agents can also send the latest address position and transaction reports via email.

•The project is publicly supported by @aeyakovenko and Solana Labs, with plans to build an AI agent similar to Siri on the iPhone on Solana Mobile.

•Griffain is built by @tonyplasencia3, a former developer of the Underdog and Blink projects on Solana.

Why is this important?

Many people currently criticize AI agents as "just chatbots" and lack practical applications. And Griffin is building infrastructure for useful AI agents on the chain.

REI (market value $153 million) - A new way to play smarter AI agents

REI is not only is an ordinary AI agent, it is an AI innovationA full-stack framework integrated with blockchain. Think of REI as an agent that not only acts, but also learns, reasons, and adapts.

Function highlights:

•REI mainly interacts with users through X (Twitter), with Natural and contextual conversational skills.

•Oracle Bridge connects off-chain data, while the ERCData standard optimizes the gas cost of on-chain data, making on-chain AI applications more efficient.

•REI’s 4-layer cognitive model (thinking, reasoning, decision-making, execution) gives agents the ability to learn and self-adjust, enabling deterministic execution.

Recent catalysts:

•AI Stack SDKs: launched in phases to reduce developer creation Threshold for agency.

•Extended oracle network: Laying the foundation for building real-world AI applications on the chain.

•Phase 2 plan: comprehensive testnet, advanced verification protocol, and 2025 mainnet plan.

What is the overall situation? REI’s goal is not just to build smarter robots, but to shape the future of decentralized agents through the precise combination of AI and blockchain. If you care about the future of autonomous agents, REI is a project worth tracking for the long term.

$VAPOR - (Market Cap: $154 million) - The first and only AI agent Launchpad on Hyperliquid

Vapor is an AI agent Launchpad built natively on Hyperliquid, using @ai16zdao's Eliza framework.

Core Takeaways:

•Although $AI16Z accounts for approximately Solana’s fully diluted valuation (FDV )of1.2%, but $VAPOR has a market cap of only $132 million, equivalent to 0.47% of Hyperliquid FDV.

•After going online, users must destroy $VAPOR to start the proxy, which means that $VAPOR is a deflationary token and the total supply will continue to decrease.

•All proxy tokens launched on Hyper EVM will be traded on a native DEX pool paired with $VAPOR. As more agents are launched, more $VAPOR will be locked in the liquidity pool.

•Already online on the testnet, all agents currently do not need to hold tokens to approve.

MODE (market cap: $114 million) - The intersection of AI agents and DeFi

Can be Mode Network is regarded as a Layer 2 of Ethereum, focusing on expanding DeFi to billions of users through on-chain AI agents.

"By 2026, more than 80% of on-chain transactions will be completed through AI agents." If you believe this prediction, Mode will let you plan ahead.

•TVL: Over $500 million, the third largest chain in the OP Superchain ecosystem.

•User growth: 367,000 users and 24 million+ transactions.

•Undervalued: MC/TVL ratio shows potential upside.

Mode’s latest news:

•Synth testnet online: Bittensor subnet, using For financial prediction, unlock the intelligence of DeFi.

•AI Agent Hackathon: The first hackathon, which attracted 667 registrants and resulted in the development of 10+ Eliza plugins in just 6 days.

•AI agent application store: There are currently multiple agents online, including:

•Amplifi: Bitcoin abstraction layer, implemented using AI Bitcoin Extensions

•Arma: Optimizing USDC Yield Farms

•Brian: Allowed. Natural language execution of transactions and contracts

•Sturdy: Aggregate AI-driven revenue vaults to increase profitability.

•QuillAI Network: Secure AI powered by D-LLM and EigenLayer AVS Proxy.

Mode's vision: Building an infrastructure for the on-chain proxy economy. With the Synth testnet, BitcoinOS integration and the growing AI proxy ecosystem, Mode Will play an important role

$GAME (market cap: $390 million) - the backbone of the Virtuals framework

GAME is the first infrastructure project to allow developers to integrate AI agents into games

Description from Virtuals:

“Every society has core infrastructure that drives its development and growth. @game_virtuals plays the role of Shopify in the Virtuals ecosystem, while @base is similar to SWIFT. ”

Notable progress:

•Developed directly by the Virtuals Protocol team.

•$GAME is the backbone of the virtual economy. Every time the AI ​​agent runs, $GAME is paid as an operating expense.

•GAME has been provided for more than 200 projectsSupport, handling over 150,000 requests per day.

•G.A.M.E has the #1 30-day performance in the Virtuals ecosystem, growing nearly 400%.

•GAME has the most Lifetime Inferences, with over 2.2 million.

Why is $GAME undervalued? Not much is known about its capabilities, and even the G.A.M.E team themselves don’t promote it heavily (the official account only has 5 tweets and doesn’t even change the Twitter profile picture).

A similar but riskier option is CONVO, which powers AI agent conversations on Virtuals.

Soul Graph ($GRPH, market cap: $36 million) - Making AI agents more humane

Imagine that if other frameworks shape the "hard skills" of the agent, then Soul Graph shapes its "soft skills", making the AI ​​agent more human-like. Its framework is designed to add personality to new or existing AI agents, giving them persistent memory and real-time communication capabilities.

"We are building open source tools and infrastructure that allow developers to give agents evolving personalities, persistent memories, and real-time communication capabilities."

Core takeaways:

•Over 2000 "souls" have been created, available in the Soul Graph Playground 15,000 minutes of voice interaction time.

•Recently integrated AI16Z plug-in.

•85% of the total token supply is allocated to the community, with the development team’s allocation locked and vested for 1 year.

The reason why I am optimistic about this project:

Character AI is very popular. ordinaryUsers spend hours a day chatting with AI, and Google just acquired Character.ai for $2.7 billion. Imagine if users could interact with an AI agent with a human personality in real time? Soul Graph is making that happen.

Other projects worthy of attention

•Bully (@dolos_diary): Originally a "complaint" account, now developed a framework for social media AI agents@ dolion_ai.

•TopHat (@tophat_one): AI agent Launchpad on Solana.

•Swarm Node (@Swarmnode): Similar to AWS Lambda, but specifically for AI agents and frameworks.

•Cookie Fun (@cookiedotfun): A data tracking platform for AI agents, similar to DeFiLlama/Kaito, tracking the agent's influence, market share, etc. Advanced features require tokens to access.

How to choose?

First of all, Virtuals and AI16z are required configurations. They are clearly dominating the market and I don’t see this trend slowing down anytime soon, the network effects cannot be underestimated.

If considering other projects with potential, I pay attention to the following points:

•New LaunchPad on the chain: This is the strategy template for 2021. Virtuals dominates Base, so you can find LaunchPad on other popular chains, such as $VAPOR, which has seen an astonishing rise on Hyperliquid.

•Market segment strategy: It is difficult to launch a general-purpose L1/L2, and there are more opportunities to focus on market segments. For example, Mode focuses on the combination of AI agents and DeFi, while Griffin focuses on natural language interaction.

•Programming languages: Infrastructure requires agents, agents require developers, and developers have their own preferred languages. Therefore, Zerebro (Python) and Arc (Rust) have strongcompetitiveness.

AI investment portfolio ideas

1. Buddhist portfolio - 50% invested in Virtuals, 50% invested in ai16z. This simple combination could outperform 80% of investors.

2. Alpha/Beta combination - use Virtuals as the core Alpha investment, and then choose competing frameworks such as Arc or Griffin as Beta options.

3. Barbell strategy - Virtuals and ai16z as "sound" bets, coupled with various high-risk A.I. agent projects.

4. Comprehensive layout - infrastructure Alpha investment, plus some Beta projects, and then diversify investment into A.I. agent projects with different risk levels. Including leading agents like @aixbt, some small projects with market capitalization below $10m can also be allocated.

Criticisms of the AI ​​agent track

"Too many Launchpads and frameworks" – Guys, have you seen how many L1 and L2 there are now? We are growing very well. This field is new and progressing rapidly. If you want to build a new city, it makes sense to build roads, bridges, and pipes first.

"Most of it is garbage and eventually goes to zero" – This is the crypto world, and most things do go to zero. But those that survive will be game-changers.

"The AI ​​track has peaked" – you are destined not to succeed. Don’t be too smart and take the middle road.

"All AI agents suck" – Remember a time when people used to be afraid of shopping online? I also remember that the first iPhone couldn't even copy and paste. We are still in the early stages and this is a bet on the future.

Conclusion

I know there will be a bull market this cycle, but I am not sure what new, game-changing primitives will emerge.

Every cycle will have a moment like "DeFi" or "NFT" - a track that revitalizes developers, a track that revitalizes retail investors Enlightenment and madnessA wildly crowded track.

I once doubted whether this moment would really come. But the good news is...it's already here. This is your chance to achieve financial freedom—don’t blow it.