CopilotKit raises $27M for agents

AI agents are moving out of the chat box and into the product itself. That is the bet behind CopilotKit’s new $27 million Series A, announced May 5, and it matters because the hard part of enterprise AI is no longer just model quality. It is the interface layer — how an agent shows its work, asks for approval, updates the screen, and stays inside company rules. CopilotKit wants to own that layer. ### What did CopilotKit actually raise? CopilotKit said it raised a $27 million Series A led by Glilot Capital, NFX, and SignalFire. The company is based in Seattle and has spent the last two years building open-source tools for developers who want AI agents to live inside existing apps instead of beside them in a generic chatbot. (copilotkit.ai) ### What problem is it trying to solve? The basic complaint is easy to understand — chatbots are clunky inside real software. If a user asks for a revenue breakdown, a wall of text is often the wrong answer. What developers actually want is an agent that can return a chart, open the right panel, ask for confirmation before taking an action, and keep the app state in sync while the user(copilotkit.ai). (techcrunch.com) ### So what is AG-UI? AG-UI is CopilotKit’s open protocol for agent-user interaction. In plain English, it is a standard way for an agent backend and a user-facing app to talk to each other in real time. It handles streaming responses, tool calls, shared state, approvals, and custom UI rendering — basically the mechanics that make an agent feel like part of the product instead of a bolt-on assistant. (copilotkit.ai) ### Why does a protocol matter here? Because the market is getting messy fast. Model Context Protocol handles how tools and context get exposed to models. Agent-to-agent protocols handle coordination between agents. AG-UI is trying to define a different boundary — the handoff between agent and user interface. If that layer fragments, every team ends up rebuilding the same glue code for streami(copilotkit.ai)owers that cost and makes frontends more portable across agent backends. (copilotkit.ai) ### Is anyone important actually using it? CopilotKit says yes — and this is the part investors are clearly buying. The company says it is used in production by the majority of the Fortune 500 and Global 50, naming DocuSign, S&P Global, Cisco, Deutsche Telekom, and Function Health. It also says AG-UI has been adopted by Google, Microsoft, Amazon, and Oracle. Microsoft’s own documentation now i(copilotkit.ai) which is a useful signal that this is not just a house protocol. (copilotkit.ai) ### What is CopilotKit selling beyond open source? The open protocol is the wedge, but the business is the enterprise stack on top. CopilotKit is packaging support, self-hosted deployment, governance, analytics, and guardrails for companies that need agents inside regulated or mission-critical workflows. That is the familiar open-source playbook — win developers first, then sell the operational layer enterprises need to ship safely. (techcrunch.com) ### Where does the risk sit? At the boundary where the agent touches the user. The more an agent can change a screen, trigger workflows, or act on live company data, the more approvals, observability, and policy controls matter. CopilotKit leans into that by talking about guardrails at the boundary, but that also tells you wher(techcrunch.com)that makes the mistake look trustworthy. (copilotkit.ai) ### Bottom line? This round is really a bet that the next AI platform fight is not just model versus model. It is interface versus interface. If agents become normal parts of business software, the winners may be the companies that make those agents legible, controllable, and safe enough to use inside the app people already trust. (copilotkit.ai)