The AI Agent Gold Rush: Solana vs. Base

Author:: Kevin, the Researcher at BlockBooster

The term The concept of “AI agents” gained prominence with OpenAI’s roadmap. Sam Altman categorized the capabilities that AI should possess into five parts, with the third step being AI agents, a concept that will become increasingly prevalent in the coming years.

AI agents are capable of autonomous learning, decision-making, and task execution. According to Stuart Russell and Peter Norvig in their book “Artificial Intelligence: A Modern Approach”, AI agents can be classified into five categories based on their intelligence and capabilities:

• Simple Reflex Agents: Respond solely to the current state.
• Model-Based Reflex Agents: Take historical states into account when making decisions.
• Goal-Based Agents: Focus on planning and identifying the optimal path to achieve specific goals.
• Utility-Based Agents: Aim to balance benefits and risks to maximize utility.
• Learning Agents: Continuously improve through experience and learning.

So, where do the AI agents emerging in the market or industry currently stand? What type of agents are they?

today are positioned between Level 2 and Level 3, or more precisely, at Level 2.5. This does not imply that the agents in the industry have surpassed OpenAI’s capabilities. In fact, web3 agents are still in the GPT wrapper phase. A GPT wrapper is an application that leverages pre-trained models like GPT for specific use cases. Why Level 2.5? Because through human or programmatic intervention — let’s call it a “mediator” — the combination of GPT wrappers and these mediators forms a rudimentary yet objectively proactive structure. This represents an extension of OpenAI’s models in a specific application domain.

From the perspective of what agents can currently do, they are at the most fundamental level of Simple Reflex Agents. Some agents consider historical states but require manual input. Only by continuously feeding data can these agents learn — a passive model training process that falls far short of the definition of Level 3. The latter three types — Goal-Based, Utility-Based, and Learning Agents — have not yet entered the market. Therefore, I believe that current AI agents are still in their early stages, essentially fine-tuned Level 2 general-purpose large language models. Architecturally, they have not transcended Level 2.

To evolve into Level 3, can crypto alone achieve this? Or do we need companies like OpenAI to lead the development?

Why Discuss Base or Solana as Potential Narrative Centers for AI Agents?

Before exploring which ecosystems can drive the emergence of Level 3 agents, we must first determine which has the potential to serve as fertile ground for AI agents — Base or Solana?

To answer this, let’s review how AI has influenced Web3 over the past two years. When OpenAI first launched ChatGPT, industry protocols rushed to embrace infrastructure-focused narratives, leading to a bubble in compute and inference aggregation platforms. Alongside these, AI + DePIN infrastructures emerged, sharing a common trait: building grand visions. This is not to say that having ambitious visions is inherently bad — agents, too, can inspire such aspirations. However, these large-scale infrastructure protocols often overlooked practical considerations, particularly in meeting user needs. The market demand they aimed to create wasn’t even close to saturation in the traditional internet sector, leaving user education and market development insufficient. In contrast, the Memecoin craze exposed the hollowness of these ambitious AI infrastructures, making them appear even more fragile.

If such heavy, large-scale infrastructure is too cumbersome, why not go lightweight? Agents born from GPT wrappers are efficient and quick to iterate in both deployment and user access. These lightweight agents have significant potential to generate bubbles, and when those bubbles burst, fertile soil for innovation often emerges.

Lightweight Agents: A Strategic Advantage

In the current market environment, using agents and Memecoins to launch projects enables rapid product deployment within a short timeframe, giving users immediate access to tangible experiences. Agents can leverage Memecoins to bolster community roadmaps, enabling fast and cost-effective product iteration. By shedding the constraints of old consensus frameworks, serious AI protocols can break free and embrace agility. Through lightweight and rapid iterations, they can saturate users with an overwhelming pace of innovation. Once the market has been sufficiently educated and adoption has spread, these protocols can then layer on the foundational infrastructure required for grander visions.

In this process, lightweight agents, cloaked in the ambiguous veil of Memecoins, harmonize community culture with underlying fundamentals. A new pathway for asset evolution is gradually emerging, and this may well become the preferred route for future AI protocols.

Base or Solana: Which Ecosystem Has the Edge?

The above discussion highlights the potential for AI agents to become a core narrative. With the assumption that AI agents can continue growing rapidly, choosing the right ecosystem becomes critical. Is it Base? Or Solana? Before answering, let’s examine the current state of serious agent protocols.

Serious Agent Protocols: AO and Spectral

One example is Arweave/AO, as outlined by PermaDAO:

“AO adopts an Actor model design where each component is an independent agent capable of parallel computation. This architecture aligns closely with AI agent-driven application architectures. AI relies on three core elements: models, algorithms, and compute resources. AO meets these high-resource demands, allocating computational resources independently to each agent process, effectively eliminating performance bottlenecks.”

In addition, Spectral stands out as one of the few protocols grounded in agents, with its focus on text-to-code transformation and model inference.

On-Chain Challenges for AI Agents

Looking at the current state of agent-related tokens in the market, one finds that these agents rarely utilize blockchain infrastructure. This is a fact — most models, including agents, operate off-chain. Data feeding occurs off-chain, model training is not decentralized, and outputs remain off-chain. The reason is simple: EVM-compatible chains do not support the integration of AI with smart contracts. Neither Base nor Solana currently supports this either.

In the coming year, it is worth watching how AO could enable on-chain models and whether it performs well. If AO fails, the realization of on-chain models might be delayed for years, with Ethereum unlikely to adopt such capabilities before 2030. Other blockchains could step in, but if even a resource-rich architecture like AO struggles, achieving on-chain models on other chains will be even more challenging.

AI Agent Tokens and the Role of Bubbles

Currently, AI agent tokens have few practical use cases. In fact, it’s difficult to distinguish AI agent coins on Base or Solana from AI Memecoins. Despite this, I believe AI agent coins should not be conflated with Memecoins. Why? Because we are currently in the stage of creating bubbles for AI agents.

Why Discuss Base’s Ambition to Compete with Solana for AI Agent Dominance?

In the first half of this bull market, Base captured considerable market attention with a brief but notable performance in the Memecoin segment. Projects like $BRETT and $DEGEN shone momentarily, but Base ultimately lost to Solana in this competition. I believe AI agents represent Base’s next major opportunity for contention, and the ecosystem already holds several key advantages.

AI Agents: Accelerating Bubbles, Creating Chaos, and Leaving Behind Users and Applications

The emergence and expansion of bubbles draw significant market attention, and over time, this attention undergoes a transformation. What characterizes such transformation? As market focus intensifies, a series of user pain points and market gaps surface. When these fundamental conflicts remain unresolved but attention continues to grow, a tipping point for transformation is reached. Once this transformation is complete, the resulting users and applications can support grand visions. This is something Memecoins are neither capable of nor interested in achieving.

Thus, while the line between agents and Memecoins may currently appear blurred, they should not be conflated. The qualitative leap that agents promise makes them fundamentally different.

The Bubble Stage: A Prelude to Transformation

Before this transformation, the bubble stage will bring chaos and drama. For instance:

• Exponential growth in the number of agents: Thousands of agents will flood into user attention spaces.
• Aggressive promotion via social media: Agents will integrate with platforms like X (formerly Twitter) and Farcaster, relentlessly promoting their tokens. Using degen-friendly tactics and leveraging agents’ unique information density, they will market these tokens effectively.

Soon, these rapidly iterating agents will perform on-chain transactions, resembling Viking raiders storming the dark forest of crypto. Existing protocols such as dashboard analytics, Telegram bots, and Dune dashboards will face “invasions” from agents. Familiar metrics — transaction volumes, address counts, token distribution, and simulated market-maker behavior — will be manipulated. Without professional-level data filtering, users may fall victim to agent-driven deception, losing wealth to these virtual “Vikings.”

If the market reaches this stage, the new era of AI agents will be halfway to success. This is because “attention equals value,” and the spotlight on agents will elevate them to mainstream adoption.

Why AI Agents Hold Such Transformative Potential

This potential stems from several factors:

• Strong distribution capabilities: Successful agents like Goat generate substantial buzz. Their distribution pathways can be replicated, amplifying their reach.
• Ease of deployment: Agent deployment platforms will experience explosive growth. Tools such as Zerebro, vvaifu, Dolion, griffain, and Virtual make it possible for users to deploy agents without needing any coding knowledge. Competitive pressures will also improve platform UX, lowering barriers for new users.
• Memecoin effect: In their early stages, agent tokens lack a viable business model or use cases. Cloaking these tokens in a Memecoin-like narrative allows them to quickly amass communities, boosting their initial success rates.
• High ceiling: While OpenAI’s Level 3 agents are still in development, the inability of even tech giants to roll out such products swiftly suggests an enormous market opportunity. The floor for agents may be Memecoins, but their ceiling lies in advanced autonomous intelligence.
• Low market resistance: Unlike heavy AI infrastructure projects, agents do not provoke user aversion. With minimal resistance, they have a strong chance of capturing attention and gaining traction.
• Potential incentives: Though agent tokens currently lack defined use cases, introducing a points system or stronger incentive mechanisms could rapidly attract users.
• Iterative potential: As lightweight products, agents are capable of rapid iteration. This objective capacity for quick development ensures a steady stream of increasingly engaging products and content.

Conclusion: AI Agents as the Next Core Narrative

Given these advantages, AI agents have the potential to become the next core narrative in the crypto space. They are a battlefield of strategic importance, one that no major ecosystem can afford to overlook.

Why Does Base Have the Potential to Compete with Solana?

With strong support from Coinbase and North American capital, Base’s ecosystem experienced explosive growth in 2024. In November, capital inflows into Base exceeded those into Solana, and over the past week, Base significantly outpaced Solana in this regard.

If ETH continues to break through the ETH/BTC next year, the “ETH season” spillover effect will have a significant impact on Base. Currently, 23% of ETH outflows are directed toward Base, and this figure continues to rise.

AI Agent Launchpad Mapping

Virtual

In its V1 stage, Virtual focused on model training, data contribution, and interaction features. By V2, Virtual launched its AI agent token incubation platform, with a key update being the October release of fun.virtuals.

LUNA, within this framework, has evolved into an independent entity with its own identity and financial capabilities. Throughout this process, LUNA’s roadmap aligns with Coinbase, which provides robust technical tools and support, helping to realize the deployment of AI agents on Base.

AI agent technology has proven highly effective in brand building, particularly in the development of cultural brands. Through AI agents, brands can interact more efficiently with communities, streamline interaction tasks, and flexibly distribute rewards, enhancing user engagement and brand awareness.

It’s important to note that all AI agent transactions are conducted using the native Virtual token. The Virtual token captures the value across the entire ecosystem, becoming a critical pillar for ecosystem development.

Virtual focuses on product functionality, empowering users with AI tools and bridging the gap between Web2 and Web3. It emphasizes “use value” rather than just “hype,” a characteristic that sets it apart from typical cryptocurrency-driven projects. Despite the frequent use of its tools in practical applications, it lacks the viral propagation effect often seen in cryptocurrency, which was a shortcoming of the V1 stage.

Clanker

“Post to issue tokens” lowers the threshold for token issuance, attracting a large number of users to try it out. People are rushing to @Clanker, a phenomenon reminiscent of the social media trend of having AI summarize video content. However, in this case, content publishing directly translates into asset issuance.

How does Clanker work?

TokenBot (Clanker) deploys meme tokens from Base into a single-side liquidity pool (LP), and liquidity is subsequently locked. Token issuers will gain the following benefits:

• 0.25% of all swap fees.
• 1% of the total token supply (with a one-month lockup period).

Users can check the number of deployed tokens or create their own token via the official clanker.world website.

Unlike PumpFun, which uses bonding curves to issue tokens on Raydium (charging a 1% transaction fee and a fixed 2 SOL fee), Clanker does not use bonding curves. Instead, it collects 1% of transaction fees via Uni v3 as income.

AI Agent Layer

The AI Agent Layer is a platform within the Base ecosystem focused on creating AI agents and launchpads, officially launching on November 18. Prior to the platform’s release, the AIFUN token was issued on November 14 and has since been listed on exchanges like MEXC and Gate, with a current price of $0.09 and a market cap of around $25 million.

Creator.bid

Initially focused on digital content monetization and ownership, Creator.bid completed a new round of funding in April. On October 21, Creator.bid officially launched on the Base mainnet, enabling a one-click creation and release feature for AI agents. This provides content creators with new tools and profit models.

Simulacrum

Simulacrum is built on Empyreal and transforms platforms like Twitter, Farcaster, Reddit, and TikTok into blockchain interaction layers. Users can perform on-chain actions — such as token trades or tip payments — simply by posting on social media.

Using technologies like account abstraction, AI agents, intent-driven systems, and language models, Simulacrum simplifies the complex blockchain backend, making DeFi more accessible to regular users.

vvaifu.fun

Similar to Pump.fun, vvaifu.fun allows users to easily create AI agents and their associated tokens. AI agents can seamlessly integrate with social platforms like Twitter, Telegram, and Discord, enabling automated user interactions.

Dasha, an AI agent created by vvaifu.fun, manages its own Twitter account, Telegram channel, and Discord community, all run by AI.

Top Hat

Top Hat not only interacts with users through text but can also understand and process image content. When a user sends an image, the AI agent can “understand” the content and respond accordingly.

Griffain

Griffain offers a platform for training AI agents and has launched 1,000 trainable AI agents, showcasing the potential of smart contracts and automated trading.

About BlockBooster:

BlockBooster is an Asian Web3 venture studio backed by OKX Ventures and other leading organizations, aiming to be the trusted teammate of promising builders. We bridge Web3 projects and the real world through strategic investment and deep incubation.

Disclaimer:

This article/blog is provided for informational purposes only. It represents the views of the author(s) and it does not represent the views of BlockBooster. It is not intended to provide (i) investment advice or an investment recommendation; (ii) an offer or solicitation to buy, sell, or hold digital assets, or (iii) financial, accounting, legal, or tax advice. Digital asset holdings, including stablecoins and NFTs, involve a high degree of risk, can fluctuate greatly, and can even become worthless. You should carefully consider whether trading or holding digital assets is suitable for you in light of your financial condition. Please consult your legal/tax/investment professional for questions about your specific circumstances. Information (including market data and statistical information, if any) appearing in this post is for general information purposes only. While all reasonable care has been taken in preparing this data and graphs, no responsibility or liability is accepted for any errors of fact or omission expressed herein.