AI agents are becoming a new and rapidly growing class of API consumers. APIs were traditionally designed for human developers and users: someone sends a request, examines the response and decides what to do next. AI agents can instead query APIs, interpret the results and immediately take another action, creating long-running machine-to-machine workflows with little or no human intervention.
This shift is particularly important in financial services, where agents are already being used for payments, claims processing, fraud detection, trade accounting and customer onboarding. The article highlights examples including Goldman Sachs using autonomous agents for trade accounting and onboarding, while Lloyds expects enterprise-wide agentic deployment to generate significant value. As agents gain the ability to move money and make decisions, identity, authorization and trust become much more important than they were for conventional software APIs.
The emerging agent economy will therefore need new infrastructure beyond APIs themselves. The article points to agent identity and discovery, reputation and trust systems, insurance and legal mechanisms for agent mistakes, and micropayment infrastructure capable of handling machine-speed transactions. This could create an entirely new technology layer in which agents discover one another, authenticate themselves, transact and coordinate without requiring humans to manually manage every interaction.
The broader takeaway is that the API economy may be evolving into an agent economy. The competitive advantage may increasingly depend not on which company has the best model, but on whether its agents can reliably orchestrate multiple services and operate across complex workflows. That means businesses will need to rethink API pricing, authentication, permissions and reliability for customers that are no longer humans or developers—but autonomous software systems acting on their behalf.
