What if artificial intelligence (AI) agents could hire other AI agents to help them with their work?

That’s essentially the idea behind Mech Marketplace, a decentralized platform from crypto-AI firm Olas that enables easy collaboration between various AI agents on an autonomous basis.

“We’re not at a point where AI agents can do everything,” David Minarsch, a founding member of Olas, said in an interview. “In order to architect the systems, it’s better if you separate concerns. So you let different agents specialize in different things, and that specialization gives rise to the need for agents to trade with each other.”

For example, if you’re using an AI agent to make money from prediction markets, the agent would typically know how to interact with the betting platform — how to place orders, how to withdraw funds — but it may need help from a different AI agent to make the actual predictions.

AI agents can already interact with each other; Olas has recorded more than four million transactions within its ecosystem, and more than half between agents. The difference, Minarsch said, is that AI agents can currently find other agents only by having their specific identities coded in.

The marketplace, then, makes things more dynamic; instead of being programmed to interact with specific bots, AI agents can now simply come to the Mech Marketplace and find whatever they need.

Olas is supported by a variety of blockchains: Ethereum, Solana, Polygon, Arbitrum, Optimism, Base and Gnosis Chain, with the vast majority of transactions occurring on this last chain. Almost 2,000 AI agents have already been deployed across Olas’ ecosystem, about 500 of which are active on a daily basis.

The firm recently raised $13.8 million to help it launch Pearl, an app store that allows users to own AI agents.

Down the line, AI agents will likely be able to abstract away some of crypto’s complexities, Minarsch said. Instead of manually seeding wallets, bridging networks, and managing DeFi yield vaults, users will simply be able to tell the agent what to do for them.

“We will find ourselves in a situation where every human will have multiple agents augmenting their daily lives and fully autonomously doing things for them,” Minarsch said.