Circle CEO: AI Agents Accelerating Product Velocity

Circle's strategic shift involves embedding AI agents not just as a feature, but as the core mechanism for work delivery, a move described as "agentic engineering." This approach uses AI to plan and execute complex, multi-step tasks like adding features or fixing bugs, fundamentally changing how software is built, verified, and maintained. The goal is to create autonomous systems that can manage the entire software development lifecycle with minimal human intervention. This transition to an agent-driven model is part of a broader vision articulated by CEO Jeremy Allaire, who foresees tens or hundreds of billions of AI agents conducting economic activity online. In this future, stablecoins like USDC become the essential medium of exchange for agent-to-agent transactions. Circle is actively contributing to standards for agentic payments, such as Google's X402, to position USDC as the default currency for this emerging machine economy. The acceleration in product development is underpinned by significant growth in Circle's core business. In the fourth quarter of 2025, Circle's on-chain USDC transaction volume neared $12 trillion, a 247% year-over-year increase. During the same period, the company's adjusted EBITDA surged by 412% to $167 million, with total revenue growing 77% to $770 million. Circle is building the infrastructure to support this vision, including Arc, an enterprise-grade blockchain designed as an "Economic Operating System" for the internet. The Arc testnet processed over 150 million transactions in its first 90 days, with settlement times averaging around half a second. This platform, combined with tools like the Cross-Chain Transfer Protocol (CCTP) which has already handled $126 billion in volume, is intended to provide a robust foundation for an internet-native financial system. The move toward agentic AI is not without challenges, as it can amplify existing organizational weaknesses in processes and quality assurance. Success depends more on disciplined implementation—clear constraints, context control, and verification—than on the capability of the AI models themselves. This shift is forcing a complete re-evaluation of the traditional software development lifecycle, moving from structured handoffs to a more dynamic and interconnected process. Gartner predicts that by 2027, 80% of software projects will utilize AI-powered coding assistants, a significant increase from less than 10% in 2023. This trend extends beyond just coding, with AI being applied to automate testing, improve decision-making through real-time data analysis, and reduce development costs. The global agentic AI market is projected to grow from approximately $5.1 billion to $47 billion by 2030. For product leaders, this represents a structural shift from feature-led to workflow-led development. Instead of merely adding AI features to existing products, the focus is on creating autonomous systems that orchestrate complex workflows across different tools and teams. This requires a new product lifecycle management approach that accounts for the experimental and data-dependent nature of AI development, from initial scoping and data preparation to continuous monitoring and retraining of models. In the fintech space specifically, AI agents are already being deployed for fraud detection, risk assessment, and automating customer support. By analyzing vast amounts of transactional and behavioral data, these agents can identify anomalies in real-time and personalize customer experiences. This allows financial institutions to improve operational efficiency, reduce human error, and offer more inclusive services like AI-driven credit scoring.