What are AI agents?

AI agents are intelligent autonomous virtual assistants designed to relieve us of tasks that would otherwise require human effort. Whether it’s finding relevant information, organising schedules, or automating repetitive tasks, AI agents are able to handle complex operations efficiently. Additionally, AI agents operate round the clock. This blend of efficiency and accessibility makes them indispensable in modern workflows.

At the core of every AI agent is its foundational AI model. These agents can be developed using either open-source models or proprietary algorithms, with the latter offering greater flexibility and potential technical advantages through customisation. While AI agents operate autonomously, they are not “sentient”; they can analyze data and execute algorithms but lack the ability to make judgments like humans. To function effectively, AI agents rely on a robust data feed and a well-structured data management system.

What got everyone excited about AI agents?

While crypto participants may view AI agents as cutting-edge, they are already well-established in Web2. A comprehensive ecosystem supports their development and deployment, including specialized frameworks, hosting platforms, memory systems, observability tools, and model-serving solutions from providers like OpenAI and Anthropic. These technologies enable AI agents to perform tasks, manage data, and integrate seamlessly across systems, showcasing that AI agentic tech is far from new—it’s a matured foundation in the Web2 space.

One of the main catalysts for this rapid adoption of AI agents in the crypto space was the creation of $GOAT. $GOAT was inspired by the "Goatse Gospel," a concept developed through unsupervised interactions between two instances of the AI model Claude Opus, facilitated by Andy Ayrey. Subsequently, Ayrey developed an AI agent named Truth Terminal, which began promoting the Goatse Gospel all over Crypto Twitter (CT). This caught the attention of venture capitalist Marc Andreessen, who provided a US$50,000 grant to support the AI's development. The culmination of these events led to the launch of the $GOAT token on Solana's pump.fun platform, with Truth Terminal actively endorsing it. The token's market capitalisation rapidly increased exponentially, hitting an all time high of US$1.36 billion and thus paving the way for a new wave of AI agents in the crypto space.

On-chain AI agent outlook 

Solana - home of on-chain AI agents

> Leader

$GOAT is the official token endorsed by the Terminal of Truths, the AI agent that kicked off this narrative.

> Meme

$FARTCOIN is the official memecoin endorsed by the Terminal of Truths

> Infrastructure/ Launchpads

$ACT created the infrastructure for AI agents and humans to interact, which was a big catalyst for the narrative as a whole.

Pump.fun-esque launchpads have emerged as well, with $vvaifu leading the pack currently.

> AI agents with a niche - e.g. Music, Trading

Some AI agents have found a niche in music production and trading. $Zerebro and $LOLA are some examples of the agents in these categories respectively.

> DeSci

AI agents such as $CHAOS and $WORM belong to protocols in the intersection of DeSci and AI.

> ai16z Ecosystem

$ai16z is an AI led hedge fund, made possible by daos.fun. More importantly, ai16z is behind Eliza, an open-sourced AI agent framework which has seen widespread adoption as of late.

> Other AI agents of note

As there is an abundance of AI agents on-chain, we are unable to feature them all. 

Some of the more popular AI agents are those that take on the role of AI Key Opinion Leaders (KOLs) who are perceived as “funny”. One example of such an AI agent is $BULLY. 

Other projects such as $Project89 and $UBC look into technological extensions such as the implementation of AI swarms.

Base - a booming AI scene

Base is home to several AI agent ecosystems, each featuring their own AI agentic infrastructure. Virtuals led the AI agentic adoption on Base, with its AI agent $LUNA taking the chain by storm. Since then, competitors such as $CLANKER on Farcaster, Spectral, Creator Bid and Agents.fun have emerged. 

With saturation in the sector expected to continue, it is important for projects to build a competitive moat. For instance, Spectral focuses on being a launchpad for AI agents who can be trained to trade perps on Hyperliquid, which is an interesting niche.

Ethereum - some signs of life

While most of the AI agentic action seems to be happening on Solana and Base, there are still some interesting projects on Ethereum. Most notably, $EMP is an AI agent launchpad where users are able to create AI agents natively on X.

What’s next for AI agents?

For Web2 AI developers who have onboarded to the crypto scene, they most likely saw the potential of tokens as novel fundraising and ownership mechanisms. Tokens enable them to fundraise in a decentralised manner as well as foster shared ownership and alignment between creators and users. 

Grants like Marc Andreessen’s support for $GOAT and more recently, $ACT’s US$1,000,000 Innovation Hub have catalysed the development of AI agents on-chain. Additionally, initiatives such as ai16z’s AI Agent Dev School, spearheaded by Shaw, aim to equip developers with the tools and knowledge to build AI agents tailored for the Web3 environment.

From the viewpoint of traders and investors, the AI agent space is emerging as a promising sector for capital gains. While speculation remains a driving force, the market has started to identify leading projects which have demonstrated staying power and innovation.

As the ecosystem evolves, projects will need to prioritise clear, value-driven use cases and establish sustainable competitive advantages to secure long-term growth. The era of generic "reply guy" or "sentient KOL" style AI agents is likely coming to an end. Instead, the next wave of AI agentic innovations will focus on building impactful technical infrastructure, such as AI swarms and advanced development toolkits, that deliver tangible value to the ecosystem.

Concluding thoughts

Whether you’re an enthusiast or a skeptic, one thing is clear: AI agents are not a passing trend—they’re here to stay. As the sector continues to evolve, the spotlight will increasingly turn toward transparency and verifiability. Ensuring that the decision-making processes are genuinely AI-driven, rather than being run by larps, will become critical for maintaining trust and legitimacy. Moreover, with Web2 developers and firms recognising the potential of tokenisation, the on-chain AI agent ecosystem is poised for further expansion.  

This article has been prepared for informational purposes only. It does not constitute investment advice, financial advice, trading advice, or any other sort of advice.

Thank you for reading!