Can investors make any money off the AI Agent projects on-chain? 

AI Agents, and artificial intelligence more broadly, dominate the water cooler banter and Slack channel chatter at work. It's all led by the usual players – most of them Big Tech names, others newcomers that made a name for themselves with the large language models led by OpenAI's ChatGPT.  The blockchain world got in on this, too. The biggest, most liquid names in the business crashed and burned last year and many, like Fetch.AI (CRYPTO: FET) are underwater. FET is down 90% from its all-time-highs reached in March 2024.

Still, "The AI agent narrative is one of the hottest in crypto right now and capital is chasing it hard, with both the excitement and the recklessness that implies," said MinChi Park, CO-founder and COO of CoinFello.

"Three names that keep coming up in conversations I’m having include the Virtuals Protocol, which is arguably the project that first kicked off the AI agent token meta," she said. "Virtuals built a vibrant ecosystem of autonomous AI agents with applications in gaming and social media, and I think this versatility is why so many people are chasing it."

Anyone can use Virtuals Protocol to build an AI agent and benefit from the economic activity it generates. 

She also mentioned Kite, an AI payments blockchain, which is building a key part of the infrastructure for blockchain agents. The token is up more than 50% in the last 12 months, going from $0.10 in March 2025 to $0.22 as of March 21, 2026. 

Kite operates in what is arguably the most exciting segment of the crossroad between AI and blockchain protocols. One of the technological reasons behind this trend is the emergence of the x402 payment standard launched by Coinbase (NASDAQ:COIN) May 2025. (Coinbase's share price has fallen since the AI rush, going from a high of $419 in July down to $197 currently.) The x402 standard ultimately allows AI agents to pay for services and data directly using stablecoins or other crypto assets. No humans needed. 

"The x402 payment stack is starting to take shape. The pace at which projects and infrastructure are launching is exponential," Galaxy research analyst Lucas Tcheyan wrote in a note in January.

AI16Z – named after the venture capital firm A16z but with no relation other than fanfare for its founder Marc Andreessen –. is another big one, pioneering the autonomous DAO. "Users buy AI16Z tokens and the agent invests the funds based on market data to grow its value for holders." Park said. AI16Z has been renamed to ElizaOS. The token's average 24-hour trading volume has fallen by around $2 million over the last year.

Traditional AI that we keep hearing about, such as Anthropic's agent Claude, can draft email, summarize and design documents. A Web3 AI agent can execute: it can swap tokens, stake Ethereum (CRYPTO: ETH) for yield, bridge funds across chains, or dollar-cost average into an asset, all based on a plain-English prompt from the users. 

"What makes this safe, and this is where the infrastructure conversation gets important, is the permission layer," Park said. "With smart account standards like ERC-4337 and ERC-7710, you can give an AI agent a very specific spending allowance, like a corporate card with strict limits."

Agentic DeFi and Automated Crypto Trades

DeFi automation is the clearest "real" use case today because digital, tradable assets like Ethereum are already on-chain; execution of trades can be permissionless; and outcomes of the trades are easily observable by the account holder.

"This is very real, and it’s happening now. We are one of the teams building it," Park said.

CoinFello’s OpenClaw skill, launching this month in partnership with MetaMask Smart Account Kit, lets a user's personal AI agent, such as your “MoltBot” created with the OpenClaw infrastructure, execute on-chain transactions based on natural language prompts, rather than complex computer language. You tell it “stake 1 ETH on Coinbase” and it handles the rest, asking for your approval before it makes a move. 

"I think these autonomous DAOs are the longer arc of this story," said Park.  "Right now agents are executing on behalf of individuals. The next frontier is agents acting as fiduciaries for on-chain organizations whether that is managing treasuries, executing governance decisions, rebalancing liquidity positions. The pieces are all in place technically. What is still being figured out is the legal and accountability layer. For instance, if an AI agent autonomously makes a bad trade using the DAO treasury, who is liable?"

On the payments side, Galaxy researchers believe adoption will continue to be uneven. "Certain use cases, particularly payments between agents and digital services, are likely to move quickly, while consumer-facing commerce will change little," report authors wrote in January. "In many cases, blockchains will operate behind the scenes, embedded within agent workflows rather than presented directly to end users."

Big Tech vs the Blockchain

When Big Tech players like Google (NASDAQ:GOOG) build an AI agent, every financial action that agent takes will route through their payment rails, their custody solutions and their terms of service. They will be the intermediary and own the ecosystem. 

For a huge segment of the market, such as anyone who has been debanked, anyone transacting across borders, anyone who simply doesn’t want Google knowing their financial behavior, that is a deal breaker. 

"Web3 AI agents offer something fundamentally different," said Park. "The agent executes on public, verifiable infrastructure that no single company controls." 

As for why blockchain AI projects have done so poorly in the market lately, It’s partly a resources problem, partly a culture problem, but the deepest issue is trust, said Park. 

"In a space where you are asking people to connect their wallets to autonomous AI, trust is the most variable and most fragile element in the equation. Web3 teams historically have been builder-first, communication-last, and that gap is especially costly here," said Park.  "The projects that will win over the next 18 months are the ones that learn to translate technical capability into plain-English and can demonstrate why their infrastructure is safe enough to trust with people's real money."

The writer of this article invests in Ethereum.  Artwork created by the author using Canva.

Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga’s reporting and has not been edited for content or accuracy.