From hackathons to trading tools and DAO governance, AI agents are redefining how we build and innovate

Crypto is in love with AI agents now. Firstly, AI agent sentiment is at an all-time high, as captured by Kaito.

Secondly, we’ve witnessed AI agents in the last few months integrated into memecoins (GOAT), social media influencers (Luna), venture funds (ai16z), launchpads (Virtual) and trading (Spectral).

Third, ETHGlobal hosted its largest hackathon post-Devcon (about 713 projects submitted), and the use of AI agents absolutely dominated the competition. I don’t know exactly how many teams used AI agents, but at least four judges pointed out its prevalence at the hackathon.

I should know, since I was part of a team that created Dongle Party, an app that leverages AI agents for token trading competitions.

With a suite of 1inch APIs under the hood, the application essentially abstracted away the need to navigate between wallets and dapps. Just query your AI agent to look at your wallet balance, send tokens or execute trades, all within the app.

A far more impressive team making use of AI agents was Industry AI, a top 10 finalist that created a gamified interface for multiple agents to work together to complete tasks. Industry AI was bloody impressive for a project completed in a few days, and its capabilities spanned the entire onchain experience, from wallet creation and launching an NFT collection to trading tokens.

Then there was also DAOGenie, another top 10 finalist that sought to streamline the thick bureaucracy of DAO governance with AI agents. AI applied to DAO governance isn’t a new idea, and it’s something that major DAOs like Maker have been experimenting with.

The absolute ease in which AI agents can be integrated to serve crypto use cases is largely thanks to Coinbase Developer Platform’s (CDP) AgentKit SDK running LLMs under the hood.

Such tools are making the intersection of AI and crypto simpler. If developers can do it in a multiday hackathon, brace yourself for what organized and well-funded teams are going to achieve.

Jesse Pollak, founder of Base, told me confidently at Devcon: “Base is right now at the center of the onchain AI economy. One big focus for us in the next couple months is the use of AI agents for messaging on the XMTP protocol within Coinbase Wallet. We see messaging as a key channel for AI agents to become useful points of reference.”

One thing that stuck out to me at the ETHGlobal hackathon was the natural blend of AI agents with the permissionless ethic of crypto building blocks.

In crypto sits an openly accessible stack of world-class tools that are processing billions of dollars in value daily. These battle-tested financial building blocks — DEXs, L1/L2s, bridges — are going out of their way to incentivize crypto devs to integrate AI agents to capture value for themselves, in ways that centralized exchanges and banks cannot or will not.

That’s not to say that Robinhood or Wells Fargo aren’t looking at AI agents, but that innovation will inevitably be slow and introduced in piecemeal ways within the permissioned confines of TradFi.

In both cases, it’s the simple profit motive at work. But only in the case of crypto are the rules of the game far more competitive and the pace of innovation moving at breakneck speeds.

That might explain part of the hype that has captured the imaginations of crypto researchers looking into how AI agents can reshape the blockchain economy and the onchain experience as we know it.

The use of AI agents is also but only one sliver of use cases applied to crypto, in addition to a whole host of other use cases such as data training (Grass), zero-knowledge machine learning (Giza), and of course decentralized compute (Akash).

Written by Donovan Choy for blockworks.co