AI Agents Become the API Economy’s Biggest New Customers

AI agents have emerged as the most rapid-growing segment of API users, no longer requiring human intervention to query endpoints, process responses, and autonomously execute sequences of actions. This shift is prompting a reevaluation of the foundational elements of the API economy, including business models, pricing structures, identity management, and trust infrastructure, all of which were initially conceived with human-centric consumption in mind.

A MIT Sloan article highlights the work of MIT Media Lab professor Ramesh Raskar, who contends that the true commercial potential lies not in developing specific AI agents but in creating the marketplaces, protocols, and services necessary for agents to function effectively on a large scale. Raskar imagines a future with billions or trillions of agents representing individuals, organizations, financial institutions, and objects, drawing an analogy to the transition from mainframe computing to the personal computer.

The article identifies four critical infrastructure areas required for this agent economy: systems for agent identification and discovery, trust and reputation services, insurance and legal protections for mistake management, and micropayment infrastructure based on stablecoins for high-speed transactions. A World Economic Forum report, co-developed with Accenture and featuring insights from over 150 senior leaders across 100 financial institutions, reveals that AI agents are already initiating payments, automating claims, and managing client onboarding for businesses.

Major financial players like Goldman Sachs are leveraging autonomous agents powered by Anthropic’s Claude for core trade accounting and client onboarding. Lloyds Banking Group plans a full-scale deployment of agentic AI systems by 2026, anticipating a £100 million value addition through fraud detection automation and back-office process optimization.

Allianz Partners has cut claims processing time from days to minutes using AI tools while maintaining human oversight.

Accenture's David Parker emphasizes that trust, rather than merely being a byproduct of good service, will become the product itself in this new AI-driven environment. The rapid adoption of agentic AI systems is evident in their current roles, such as initiating payments, automating claims, and managing client onboarding. For instance, Goldman Sachs is utilizing autonomous agents for core trade accounting and client onboarding, while Lloyds Banking Group aims to have these systems in place by 2026, expecting them to generate £100 million in value through fraud detection automation and back-office workflow improvements.

Allianz Partners has significantly reduced claims processing time from days to minutes with AI assistance, retaining human oversight to ensure oversight. As AI agents increasingly undertake autonomous actions on behalf of customers—from handling deposits to executing transactions and making financial decisions—the institutions that will succeed are those customers trust most to act in their best interests, rather than those with the most sophisticated models.

This transition is not just expected but is already underway, necessitating a rethinking of the API economy's infrastructure to accommodate these advanced agents.