24 Mar 2026
1h 17m

Illia Polosukhin: Why AI Agents Are Still Useless (And What Fixes Them) | NEAR Founder on IronClaw

Podcast cover

Bankless

The discussion centers on the convergence of AI and blockchain technology, particularly regarding data privacy and user empowerment. Illia Pulisukhin, co-founder of NEAR and co-author of the Transformer paper, shares his insights on how blockchain can serve as the backend for AI, providing a root of trust for identity management, marketplaces, and upgrades. He introduces IronClaw, a system designed to secure AI agents by encrypting credentials and attaching policies to prevent data breaches. The conversation explores the limitations of current AI agents, emphasizing the need for increased context and trust, and envisions a future where AI agents operate autonomously within a governed framework, potentially forming "autonomous businesses" with specific missions.

Outlines

Part 1: Security Challenges and the NEAR Origin

Part 2: Blockchain as AI Infrastructure

Part 3: IronClaw and Agent Security

Part 4: Privacy and Self-Sovereign AI

Part 5: Context, Memory, and Workflows

Part 6: Future of Work and Autonomous Agents

Part 7: Governance, Philosophy, and Advice

Sign in to continue reading, translating and more.

Open full episode in Podwise