Circle, the company behind the USDC stablecoin, now has 90% of its workforce actively using AI coding tools and desktop applications on a weekly basis. That figure comes alongside over 900 internal AI applications deployed across the organization, a number that’s grown rapidly in a very short window.
CEO Jeremy Allaire shared the milestone publicly, framing it as part of a broader effort to embed AI deeply into how the company operates.
From 85% to 90% in a month
The trajectory here is worth noting. In May 2026, Circle reported 85% weekly active users across its AI tooling, with more than 600 internal AI apps already built. One month later, adoption climbed to 90% and the app count jumped by 50% to surpass 900.
Here’s the detail that makes this more interesting than a typical corporate tech adoption story: 54% of those applications were created by non-technical employees. In English: more than half the AI tools running inside Circle were built by people who aren’t software engineers.
The Agent Stack connection
This internal AI push doesn’t exist in isolation. On May 11, 2026, Circle launched what it calls the Agent Stack, a suite of tools designed to let AI agents hold and transact with USDC autonomously.
Allaire has described AI agents as offering “new superpowers” to employees. The Agent Stack launch and the internal adoption numbers tell a consistent story: Circle is simultaneously building AI infrastructure for external developers and dogfooding those capabilities internally at scale.
What this means for investors
Circle’s internal numbers also matter for operational efficiency. A company where 90% of employees use AI tools weekly and non-engineers build their own applications is, in theory, getting more output per employee.
Circle isn’t the only company exploring AI agents in financial services. Firms like Bernstein have expressed interest in the broader AI-finance convergence. But Circle has a structural advantage that most fintech companies don’t: it controls the money rails, and Circle is building the infrastructure to make that connection seamless.
The risk, as always with corporate AI adoption metrics, is that usage doesn’t automatically equal value creation. Having 900 internal apps sounds impressive, but the quality distribution matters enormously. The 54% non-technical builder stat suggests something deeper than just deploying ChatGPT licenses, but Circle, sitting on top of billions in stablecoin transaction data, has raw material that most competitors simply don’t possess.