TLDR:
- Ethereum co-founder Joseph Lubin predicts massive agentic activity on-chain before year-end.
- MetaMask’s delegation-based agent wallet limits AI to strict user-defined policies and boundaries.
- The s402 protocol enables autonomous machine-to-machine payments without requiring human approval.
- ERC 8004 introduces on-chain agent registration and reputation tracking for accountable AI activity.
AI agents on-chain are approaching a turning point, according to Ethereum co-founder Joseph Lubin. Lubin has projected that agentic activity on blockchain networks will surge before year-end, framing the development as a structural shift rather than a trend.
His remarks come as developers race to build the infrastructure needed to support machine-driven transactions, automated wallets, and agent-to-agent payment systems across decentralized networks.
Lubin Points to Trust as the Central Barrier
Joseph Lubin has identified trust as the primary obstacle slowing AI agent adoption on-chain. Users remain hesitant to delegate tasks like portfolio rebalancing or programmatic trading to automated systems.
The fear centers on AI models producing convincing but inaccurate outputs, especially when users cannot verify the results themselves.
Lubin’s framing puts the problem in practical terms. Current AI models are stochastic by nature, meaning their outputs carry inherent uncertainty.
That unpredictability becomes a serious risk when the task involves moving real funds across a decentralized network with no intermediary to catch errors.
To address this, developers have introduced the MetaMask agent wallet, built on a delegation framework. The wallet allows AI agents to operate exclusively within boundaries set by the user. An agent cannot act outside those defined allowances, regardless of what the model concludes on its own.
This design reflects Lubin’s broader argument that human-defined policies must anchor machine behavior. Without that structure, agentic activity on-chain cannot scale beyond early adopters willing to absorb the risk of unreliable automation.
New Infrastructure Prepares the Ground for Lubin’s Prediction
Supporting Lubin’s year-end projection is a wave of protocol development targeting machine-to-machine interaction.
Lubin stated, “By the end of the year, I think we’ll see a huge amount of agentic activity on-chain,” pointing to emerging tooling as the catalyst.
The s402 protocol is built specifically for efficient payments between autonomous agents, removing the need for human approval at each transaction step.
ERC 8004 introduces on-chain agent registration and reputation tracking. Before engaging with an AI agent, other participants can review its verified history and identity. This creates a layer of accountability that has been absent from most agentic deployments to date.
Traditional finance is also accelerating its move onto the blockchain. Regulatory shifts and demand for continuous market access are pushing institutions toward new settlement infrastructure and institutional-grade wallets.
That migration adds depth and credibility to the environment where AI agents are expected to operate.
Lubin has described the emerging economy as a hybrid between human and machine intelligence, guided by human values.
As s402, ERC 8004, and policy-bound wallets mature together, the conditions he outlined for a surge in agentic activity are taking shape. His prediction increasingly reads less like a forecast and more like a timeline already in motion.