Is the Agentic Payments Wave Leaving Banks Behind?

Why It Matters

The shift lets card networks and fintechs capture future transaction volume, while banks risk losing direct customer relationships and deposit liquidity if AI agents bypass traditional banking apps.

Key Takeaways

• Visa integrates OpenAI, enabling tokenized payments via LLM agents.
• Mastercard launches Agent Pay for Machines, letting software transact with software.
• Coinbase and Robinhood allow agents to manage trades and payments.
• 78% of Americans use AI daily; 55% use it for finances.
• Banks risk losing customers to AI agents without LLM integration.

Pulse Analysis

Agentic commerce is moving from concept to reality as the world’s largest payment processors embed AI‑driven transaction capabilities into their networks. Visa’s partnership with OpenAI injects the Visa Intelligent Commerce protocol directly into large‑language models, allowing an AI assistant to initiate tokenized payments without leaving the chat interface. Mastercard’s Agent Pay for Machines, built with partners like Stripe and Solana, extends the same idea to software‑to‑software commerce, promising ultra‑high‑volume, low‑value transactions that could underpin new AI‑powered business models.

For traditional banks, the rapid rollout presents a strategic dilemma. By enabling payments outside the bank’s own app, card networks and fintechs risk disintermediating the core customer relationship that banks have cultivated for decades. Analysts cite a “customer relationship disintermediation” risk, noting that consumers may default to their preferred LLM—such as ChatGPT or Claude—to handle purchases, transfers, and even investment decisions. Without a seamless LLM integration layer, banks could be relegated to a passive ledger function, losing both fee income and deposit liquidity as fintech agents siphon activity to their platforms.

Fintechs are already capitalising on the trend. Coinbase and Robinhood have launched “agent‑first” offerings that let users delegate trading, cash management, and bill‑pay to AI agents, effectively turning the agent into a new financial interface. Consumer adoption is accelerating: the TD Bank 2026 AI Insights Report shows 78% of Americans use AI daily, and 55% rely on it for financial management—a ten‑fold increase in just one year. As agents become the default conduit for money, banks that quickly embed LLM capabilities or partner with AI platforms will preserve relevance, while laggards risk becoming obsolete back‑office processors.