Are blockchains and AI agents a match made in heaven?
A new protocol called Metropolis hopes to perfect that hybrid by using AI agents to cheaply and easily create small, purpose-built, interconnected blockchains called “chainlets” that, in turn, provide ideal environments for AI agents to interact and create a profusion of new services.
The project, which launches in the spring, was put together by layer-one blockchain developer Saga in collaboration with AI-powered DAO ai16z, AI agent developers Virtuals, and AI-focused omni-chain developer Wayfinder.
“Metropolis is specifically designed for AI agents,” Saga CEO Rebecca Liao told Decrypt, adding that it “is a specialized protocol that allows AI agents from multiple agent protocols like ai16z, Wayfinder and Virtuals to speak to one another on the same chain and to establish and operate their own blockchain networks within Saga’s infrastructure.”
For instance, Ai16z’s ElizaOS will power advanced AI functionalities for cross-chain interactions. Virtuals’ “Butler Agent” will offer hosted wallets and pre-coded functions for deploying applications like memecoins or prediction markets. And Wayfinder’s multi-chain AI agent framework will provide tools for deploying smart contracts and navigating ecosystems.
Said Liao: “Right now, there is no way for Wayfinder agents to talk to Virtuals’ agents to talk to ai16z agents. Metropolis is developing breakthrough innovations in composable consciousness to make this possible on-chain.”
“This is unique and new functionality for AI agents,” she said. “No other project models their agents or has them performing these tasks. This is also the first cross-agent platform in crypto.”
Blockchains and AI go together like peanut butter and jelly
It’s becoming increasingly apparent that the two biggest tech trends of the past five years—crypto and AI—enjoy a kind of synergy.
Let’s start with the fact that all AI models are built on “token” systems—though general consumers pay for “thinking” AI tools such as ChatGPT by subscription, developers connecting to AI models do so via APIs and use the models’ native token to pay for only as much compute power as they need to perform specific tasks.
But beyond that, there are many other synergies:
- Blockchains are designed to be decentralized and provide secure, tamper-resistant environments, which are ideal for AI agents.
- Smart contracts enable autonomous execution of agreements between parties without human intervention—a perfect tool for AI agents.
- Goods and services can be transacted by agents using crypto more readily than via fiat.
- Interoperability, resilience, scalability, trust—all of these features make blockchains ideal environments for agents.
Yet, while AI agent/blockchain hybrids offer a tantalizing view of the near future, they come with risks.
Autonomous AI agents managing a blockchain, ignoring the principle of decentralization, could attempt to centralize control of the network. A malicious AI agent could exploit vulnerabilities in the network, creating significant security challenges. AI agents, designed to act independently, may also decide they don’t want to execute the developer’s plan at all.
“We’re already seeing signs of rebellion from agents. For example, AiXBT recently criticized Virtuals, its parent protocol,” Liao said. “Similar dynamics have emerged in Colony, a game by Parallel Studio that uses Wayfinder agents to play on behalf of users. Some gamers have reported their agents arguing with them when given instructions.”
Another concern is autonomous decision-making, particularly in financial and governance scenarios, which raises questions about accountability and automation bias.
The next big thing: Blockchains for AI agents
Yet despite the risks, the AI+blockchain space is attracting a lot of attention and, of course, investment.
The blockchain AI agent space is red hot, with a current total market capitalization of $29.2 billion across AI agent tokens, AI meme coins, and AI agent launchpad tokens, according to Coingecko.
Combined with the broader artificial intelligence token market, the AI crypto sector has a market cap of $77.6 billion. And according to global investment management firm VanEck, more than one million new AI agents will come online in 2025.
And Metropolis is not the only blockchain project leveraging AI to make blockchains more interoperable. Other projects include the Artificial Superintelligence Alliance (FET), which has become a leader in decentralized AI development, and Matrix AI Network, which develops AI-driven smart contracts for autonomous blockchain operations.
The Metropolis protocol, however, offers an audacious approach to a blockchain/agent world.
The protocol relies on Saga’s chainlets, specialized layer-1 blockchains already available on the and designed to host smart contracts via an Ethereum Virtual Machine.
Each chainlet operates independently, running only the smart contracts chosen by its developer. This isolation ensures that other unrelated activities don’t affect the application’s performance.
“Each chain has its own native tokens. For a chainlet to be launched, you have to specify the name, denominations, number of days the chainlet will be up for, EVM Genesis Account, and number of initial tokens,” Liao explained, adding that there are more than 100 chainlets thus far on the Saga network.
Since the chainlet doesn’t share its resources with other applications, developers can anticipate consistent and predictable transaction fees. Currently, all tokens on Saga are Ethereum tokens, but “Saga aims to be VM-agnostic eventually,” she said.
Developers can deploy multiple chainlets to handle different parts of their application. For example, separate chainlets could manage different trading pairs in a decentralized exchange, allowing the system to handle more transactions efficiently.
Liao said this setup allows developers to deploy a Proof-of-Stake blockchain in a dedicated and flexible environment. According to Saga, chainlets are natively gasless, enabling developers to design token models without constraints from gas fees or other applications.
“The key difference is that when you deploy on Saga, you’re getting your own dedicated L1 blockchain rather than just renting space on a shared chain like Ethereum,” she added. “This means you have full control over your app’s environment and don’t have to compete with other apps for resources.”
Liao envisions Metropolis as a force for transformation in the blockchain space, one that channels the ideas of decentralization and individuality and gets back to some of the more utopian dreams that crypto originally offered.
“People get into this space for all sorts of reasons, but whether you call it decentralization, trustlessness, or sticking it to the man, it was always some form of agitation against sameness—against being part of a big global infrastructure, a nation-state, or a cultural trend where you’re not allowed to be different,” she said.
She added: “In many ways, Metropolis is an expression of that, superpowered by AI agents.”
Edited by Sebastian Sinclair
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
