27 January 2025
Why AI Agents Need Interop
The days of single-chain agents are over. To keep up with the expanding multi-chain world, AI agents need to seamlessly function across different chains. They need interop.
Agents Are Taking Over (The Timeline)
AI Agents have been the talk of the town across Crypto Twitter over the past few months. As launchpads like @virtuals_io made it easier than ever to deploy agents, they have exploded in popularity and mindshare, and everyone from degens, VCs, researchers, and institutions have taken interest in this new phenomena.
Agents aren’t complicated. They’re programs which receive a task, and then take the necessary steps to complete that task autonomously, or without human intervention. Onchain agents take this concept further with access to a crypto wallet and onchain read/write access– meaning they can retrieve onchain data and execute transactions on their own.
The excitement around agents has culminated around the idea of an “agentic economy” — a future in which onchain activity is dominated by agents. While we don’t believe this idea is implausible, it’s simply naive to think agent activity will be limited to just 1 or 2 chains.
Why Do Agents Need Interoperability?
Today, agents are primarily designed to live on a single chain. This is simply not a viable long-term solution. The rollup-centric roadmap has been underway on Ethereum for years, and its most successful L2s now have their own frameworks for building L2s. A new class of chains are taking the SVM beyond Solana. New chains are being built on Move.
In an expanding multichain world, interop allows agents to simply do more.
More Users
This is an easy starting point.
Today’s onchain economy is heavily fragmented- users and applications are dispersed across different chains and ecosystems. In order to reach as many users as possible, agents need to be able to get to as many chains as possible.
While compelling products or assets can prompt users to bridge to a specific chain (e.g., Solana’s inflows following the $TRUMP launch last week), it’s far more efficient to meet users where they already are. Agents that are accessible across multiple chains can reach more users and applications than those limited to a single chain, and are therefore better positioned for user adoption.
More Resources
It’s pretty simple: more chains = more resources for agents.
Whether it’s a thriving DeFi ecosystem, a beloved game, or a viral memecoin, each chain brings something different to the table, hence why crypto-native users rarely stick to just one chain nowadays.
Beyond chains, this concept also extends to the VMs they are built on top of. As developers continue to push the boundaries of what we build onchain, performance requirements will continue to shift. Whether its higher throughput, broader programming language support, or any other specializations for app-specific workloads, altVMs will play a crucial role in enabling the next generation of onchain applications. And cross-VM interop will become even more important.
If agents become our personal assistants one day, we’ll want them to have access to the latest cutting-edge products and services wherever they’re built. To do this, they’ll need interop.
More Blockspace
It’s important to remember that agents can operate on a MUCH greater scale than us humans do. Agents don’t care how many times they need to hit ‘approve’ and ‘sign’ to complete a transaction- they just do it.
This means that in an agentic economy, there will be A LOT more onchain activity than we see today. But blockspace is a finite resource, and as agents continue to drive more onchain activity, the workload they generate will grow exponentially. No single chain will be able to sustain this immense demand. Not without experiencing degraded performance, surging gas fees, or halting altogether, anyways.
By being able to access multiple chains, agents remain resilient. If performance on one chain falters and an agent can’t complete its tasks, it may be able to do so on another chain. Interop ensures agents can have the blockspace they need, and that our favorite chains can manage this workload.
To summarize, interop empowers agents to reach more users, capitalize on the unique strengths of different chains and VMs, and scale with the growing demands of onchain activity.
Cross-Chain Agents Today
Several teams are already ahead of the curve, pioneering solutions for taking agents across multiple chains. Some of the most notable examples include:
- Cod3x is an AI-driven DeFi platform that allows users to build custom agents in minutes with no coding required. Cod3x currently offers multi-chain support across EVM, Solana, Bitcoin, and Cosmos-based chains. Big Tony, their in-house agent, has already been trading on Base for the past 2 months with his own wallet.
- The GOAT SDK (Greatest Onchain Agent Toolkit) an open-source framework from Crossmint that allows developers to integrate various features and functionality into AI agents. The framework currently supports EVM chains and Solana, with more expansion planned ahead. @0xVictorP recently demonstrated how, using the GOAT SDK, AI agents can be used to help users bridge tokens across chains. See the demo here.
- Sympson AI is a multichain DeFi trading agent from Symphony. During the “first AI agent cross-chain transaction”, Sympson used his ETH on Base to trade $BTC perps on Gains Network on Arbitrum. More details can be read here.
- Autonolas is a developer platform for building autonomous agent systems, made up of the Olas Stack and the Olas Protocol. The Olas Stack is an open framework for creating, deploying, and managing these systems, while the Olas Protocol is a multichain protocol for coordinating and incentivizing activity between them.
- Agent Market from Holoworld AI is a launchpad on Solana for autonomous agents that can talk, act, and engage across different platforms. Agent Studio, a core part of the Agent Market, is accessible from platforms across different chains including Virtuals on Base and Heurist on the ZK Stack.
Calling All Builders
There’s currently $5000 up for grabs for the best integration of the ElizaOS framework into @hyperlane. See the bounty ⏩ here.
As we continue to build towards an agentic economy, it’s important that the right foundational rails are in place. If agents start driving more activity and managing more funds onchain, it’s critical to ensure they are trustless and verifiable. Relying on closed-source, and permissioned infrastructure is fallible.
The good news is that much of the early adoption and development around agents is being built with open frameworks.
As a permissionless, open-source interop framework, Hyperlane is a natural fit for taking the agentic economy cross-chain. To encourage innovation, there’s now an opportunity to build a Hyperlane plugin into the ElizaOS framework:
Stay tuned for more of these bounties in the near future! Or feel free to propose bounties for agent integrations!
Agents Are Here To Stay
It’s becoming clear that agents will play an integral role in the future of the onchain economy. Today, we’re seeing agents creating content, trading, and even ordering pizza from Twitter. But this is only the tip of the iceberg: the future of the agentic economy is still a largely uncharted territory, and we think it’s quite an exciting one to think about.
One thing, however, is certain: the days of single-chain agents are over. In order to unlock a flourishing agentic economy, agents need interop.
- Bring ElizaOS to Hyperlane ⏩ https://github.com/hyperlane-xyz/hyperlane-monorepo/issues/5176
More about Hyperlane
Hyperlane is the open interoperability framework. It empowers developers to connect anywhere onchain and build applications that can easily and securely communicate between multiple blockchains. Importantly, Hyperlane is fully open-source and always permissionless to build with.
