Eliza Labs Founder Warns Against Entrusting AI Agents with Large Sums

Cryptocurrency users should not trust AI agents with large sums.

Cryptocurrency users should refrain from entrusting AI agents with large sums and expecting profits, according to Eliza Labs founder Shaw Walters. He shared this view at the Token2049 conference in Singapore during a conversation with Decrypt.

He believes that in the near future, the value of the technology will lie not in capital management but in structuring signals and speeding up trade execution.

Currently, agents function best as interfaces for analytical tools and as a layer for collecting social data, he added.

In January, Eliza Labs introduced ElizaOS—an open platform on the Solana blockchain designed for creating and managing AI agents and simulations.

In the same month, the company rebranded—the project was initially called ai16z (AI16Z).

According to Walters, agents embedded in Telegram groups can anticipate market sentiment before it manifests on X. They employ simple rules, such as purchasing small amounts of tokens within minutes of a major influencer’s post.

“We can identify who makes accurate forecasts and who is trying to deceive,” the entrepreneur emphasized.

In April, Eliza Labs launched the auto․fun platform for creating, launching, and monetizing AI agents.

Issues with X

In August, the project filed a lawsuit against Elon Musk’s X Corp. The company was accused of abusing its “monopoly position” and engaging in unfair competition.

According to the lawsuit, the firm obtained key technical information from the startup, released similar AI products, and then blocked the project’s and its founder Walters’ accounts on the social network. The freeze occurred without warning or explanation.

“I just want my damn account back,” Walters said.

Development

At this stage, Eliza Labs is testing an over-the-counter platform managed by agents. It facilitates token purchase negotiations with bots within pre-set limits.

Beyond trading, Walters is interested in DAOs. In his view, they face “coordination problems” that AI agents can alleviate by summarizing discussions, bringing key decisions to the forefront, and automating workflows.

“It’s more about recommendations, analytics, and helping coordinate the community than blindly following artificial intelligence,” he emphasized.

As reported in May, Coinbase introduced the x402 tool for online payments. It is designed for transferring stablecoins via standard internet protocols and enables AI agents to conduct autonomous transactions.