Eliza Labs Founder on Why AI Agents Shouldn’t Manage Your Money—Yet

In brief

Eliza Lab’s Walters said fully autonomous trading agents are not investment-ready and perform best as interfaces that structure unstructured data into executable actions.
Eliza is building a “marketplace of trust” that paper-buys shilled tokens to score users’ calls, and an agent-run OTC desk with smart-contract guardrails.
He framed the token migration and cloud rollout as enabling governance features, multi-chain reach, and one-command agent deployment.

During an interview with Decrypt at Token2049 in Singapore on Wednesday, Eliza Labs founder Shaw Walters pushed back on the hype surrounding autonomous, profit-seeking AI traders.

Instead, he argued that the near-term value lies in structuring signals and speeding up execution rather than handing agents a treasury.

“You probably do not want to give an AI agent a bunch of money and expect it to make you more,” Walters said, adding that current agents work best as interfaces to quantitative tools and as ingestion layers for social data.

In January, Eliza Labs released the ElizaOS, an open-source platform on the Solana blockchain designed for building and managing AI agents and simulations.

Eliza’s approach includes a “marketplace of trust” that converts loose shill posts into paper trades to track results and build credibility scores.

“We can identify who’s actually good at calling and who’s trying to scam,” Walters said.

Walters also said agents embedded in Telegram groups can front-run sentiment before it surfaces on X, using simple rules such as buying small amounts within minutes of a major KOL, or key opinion leader, post.

Eliza is also piloting an agent-run OTC desk that allows users to negotiate token purchases with bots within predefined limits.

“If you can scam the agent, good for you, you had fun, but there are guardrails,” he said, pointing to caps enforced by smart contracts.

Beyond trading, Walters argued that DAOs face “coordination failures” that agents can mitigate by summarizing discussions, surfacing decisions, and automating workflows.

“It’s more about recommendation and insight and helping with community coordination than blindly following the AI,” he said.

He also took aim at Elon Musk’s X Corp, which, in August, Walters and Eliza Labs sued the social media giant, alleging the company had tricked them into handing over technical details about their AI tools, then banned them from the platform and launched copycat products.

“I just want my freaking account back,” he said, while reiterating Eliza’s commitment to open source and a multi-chain, agent-first roadmap.