Agentic Payments and the Coming AI Protocol Shift

New payment models are emerging as artificial intelligence systems move from simple assistance toward autonomous decision-making. At Sibos 2026, Coinbase’s Yuga Cohler outlined how agentic protocols could let software agents transact on behalf of users, businesses and institutions. The discussion points to a broader shift in financial infrastructure, where machine-to-machine commerce becomes a practical design goal rather than a distant concept.

The promise of agentic payments rests on protocols that can verify intent, manage permissions and settle value securely. Without trusted rails, autonomous agents could introduce new risks around fraud, accountability and operational control. Cohler’s remarks, delivered with AWS collaboration in mind, suggest that secure infrastructure will be central to making these interactions reliable across the financial ecosystem.

Banks, fintechs and crypto firms are watching this space closely. Agentic systems could streamline recurring payments, treasury operations, cross-border transfers and other workflows that still depend on manual approvals. They might also create new services for consumers, such as AI assistants that compare, negotiate and execute transactions under predefined rules.

The path forward is not purely technical. Regulators, payment networks and technology providers will need shared standards for identity, authorization and dispute resolution. If those pieces come together, AI protocols could transform finance by making payments faster, more programmable and increasingly autonomous, while raising fresh questions about oversight and trust.