Every financial API was built for a human developer, a human analyst, or a human customer. A person queries the endpoint. A person reviews the response. A person decides what to do next. That assumption is no longer reliable. AI agents are now the fastest-growing class of API consumers, querying endpoints, processing responses and taking action in extended sequences without a human in the loop. The business model, the pricing, the identity and the trust infrastructure underneath the API economy were not designed for this. They are being rebuilt now.<\/p>\n
A MIT Sloan article<\/a> profiles MIT Media Lab associate professor Ramesh Raskar, who argues that the real business opportunity in the agent economy is not building agents for specific tasks. It is building the marketplaces, protocols and services that agents will require to operate at scale.<\/p>\n The article says Raskar envisions a world of billions or trillions of agents representing people, organizations, financial institutions and objects. Every bank, every stock, every IPO will have its own agent. He compared this to the shift from mainframe computing to the personal computer. \u201cRight now, we are in the mainframe era of AI,\u201d Raskar told MIT Sloan. \u201cWe are moving to the PC era of AI, and that changes everything.\u201d<\/p>\n The article identifies four infrastructure categories this agent economy will require: agent identity and discovery systems; trust and reputation services; insurance, repair and legal services for when agents make mistakes; and stablecoin-based micropayment infrastructure for machine-speed transactions. Raskar leads MIT\u2019s Project NANDA, working to keep the agent web open before corporate consolidation locks it down. \u201cThe window to keep this web of agents open is closing soon. But it\u2019s not closed yet.\u201d<\/p>\n What Financial Institutions Are Actually Deploying<\/p>\n A WEF report<\/a>, developed with Accenture and drawing on insights from more than 150 senior leaders across 100 financial institutions, found that agentic AI systems are already initiating payments, automating claims and managing client onboarding on behalf of businesses.<\/p>\n Goldman Sachs is deploying autonomous agents powered by Anthropic\u2019s Claude to handle core trade accounting and client onboarding. Lloyds Banking Group has committed to enterprise-wide agentic deployment in 2026, expecting the systems to add \u00a3100 million in value by automating fraud detection and back-office workflows. Allianz Partners reduced claims processing time from days to minutes using an AI claims tool while keeping humans in oversight.<\/p>\n \u201cAs AI agents begin to act autonomously on behalf of customers, sweeping deposits, executing transactions and making financial decisions, the institutions that win won\u2019t be those with the most advanced models, but those customers trust most to act in their interests,\u201d said David Parker, Global Industry Lead for Banking and Capital Markets at Accenture. \u201cTrust is no longer a byproduct of good service. It is the product.\u201d<\/p>\n How Agentic Inference Is Reshaping API Consumption<\/p>\n