CopilotKit Raises $27 Million Series A

CopilotKit, the enterprise agentic frontend stack, led by Atai Barkai, Uli Barkai, and the team, has raised a $27 Million Series A funding led by Glilot Capital, NfX, and SignalFire, with participation from Discovery Ventures, Vermilion Cliffs Ventures, DVC, Abstraction Capital, 97212 Ventures, Deep Acre, J-Ventures, Gurtin Ventures, and Fresh Fund.

CopilotKit is now used by a majority of Fortune 500 companies, powering millions of daily agent-user interactions in production. Its open standard, AG-UI, has been adopted by Google, Microsoft, Amazon, and Oracle, alongside leading agent-focused startups.

The shift toward AI-mediated interfaces reflects a broader belief that all user interactions are evolving into agent-driven experiences. This funding supports the continued development of an enterprise agentic frontend stack for generative UI and agent-human collaboration.

Since its launch two years ago, CopilotKit has bet on agent-human collaboration in software, a vision that has now moved mainstream. Its AG-UI protocol, launched a year ago, has surpassed 4 million weekly downloads and is adopted by companies like Google, Amazon, Microsoft, and Oracle.

Today, CopilotKit is used by major enterprises, including DocuSign, S&P Global, Cisco, and Deutsche Telekom, powering millions of interactions weekly, with strong early growth and adoption.

What CopilotKit is solving

AI is reshaping how humans interact with software, with human intelligence remaining consistent and reliable while AI delivers highly specialized, rapidly improving capabilities. The critical layer connecting the two is where the next generation of software is being built.

Tools like Claude Code and Cursor illustrate this shift in development, with similar transformations emerging across every vertical and workflow. As AI capabilities expand, the agent–human collaboration layer becomes increasingly central, forming the foundation CopilotKit is designed to build.

Where CopilotKit is Going

Generative UI everywhere: A unified stack supporting the full range of agent-driven interfaces, from controlled UI via AG-UI to declarative UI through Google’s A2UI and open-ended experiences via Anthropic’s MCP Apps.

Self-improving agents: Building feedback loops where every interaction, accepted, rejected, or modified, enhances agent performance in real time through continuous learning, without traditional fine-tuning cycles.

AG-UI as the standard: Expanding AG-UI as the open protocol connecting agents to applications, with broader client support, deeper framework adoption, and improved tooling, positioned as an open, independent layer for the agentic web.

CopilotKit has introduced Enterprise Intelligence, a platform layer that powers persistent threads and cross-device synchronization, with observability and self-improving capabilities planned next.

Designed for production use, it is currently self-hosted on Kubernetes, with a managed cloud offering in development, enabling teams to transition agents from demos to real-world deployment.