CopilotKit nabs $27M to help developers deploy app-native AI agents

Many companies today rely on a standard chatbot interface for AI, where users type or speak a request and receive a text-based response. While functional, this approach often feels awkward and inefficient, particularly in scenarios like booking a complex travel itinerary through an app, where users must sift through lengthy text blocks.

CopilotKit’s founders, Atai Barkai and Uli Barkai, argue this method fails to leverage the full potential of AI agents and large language models. Their vision is to embed agents directly inside applications, allowing them to understand user context, execute actions, and display interactive user interfaces rather than just text. The key enabler is their open-source AG-UI protocol, which standardizes how AI agents connect to and communicate with interfaces like web browsers or apps. It supports features like streaming chat, front-end tool calls, and state sharing for human-in-the-loop workflows, giving developers a framework to deploy app-native AI agents seamlessly.

Building on AG-UI, CopilotKit is developing an enterprise toolkit that adds support, self-hosted deployment, and other business-critical features. To bring this to market, the Seattle-based startup has secured $27 million in Series A funding, led by Glilot Capital, NFX, and SignalFire.

A standout feature is the flexible user interface. CEO Atai Barkai explained that developers can provide specifications and building blocks for dynamic UIs, which the AI agent then generates contextually. “The agent can reply to you, not just with blocks of text, but with interactive UIs that are defined by your own company,” he said. For instance, a request for revenue breakdown by category could yield an interactive pie chart instead of a dense paragraph. The toolkit also gives developers granular control over how much the AI can modify the UI, from pixel-perfect designs to broad, composable components.

The funding follows strong adoption of both AG-UI and CopilotKit. The protocol, compatible with the Model Context Protocol (MCP) and Agent2Agent (A2A), is now supported by major infrastructure providers like Google, Microsoft, Amazon, and Oracle, as well as frameworks such as LangChain, Mastra, PydanticAI, and Agno. Atai noted that AG-UI sees millions of weekly installs, and a large portion of Fortune 500 companies use the protocol in production. Enterprise customers include Deutsche Telekom, Docusign, Cisco, and S&P Global.

To capitalize on this momentum, CopilotKit is launching CopilotKit Enterprise Intelligence, a self-hostable offering bundling infrastructure features for full in-app agent deployment.

Competition is fierce. Vercel’s open-source AI SDK, Assistant-ui, and OpenAI’s Apps SDK all offer similar capabilities. However, Atai argues CopilotKit differentiates itself with a horizontal, enterprise-friendly approach that supports whatever agent framework, cloud provider, or backend a company already uses. “Enterprises want optionality and self-hosting,” he said. “They’re already using Google, Amazon, Oracle, Microsoft, LangChain, Mastra stacks. They want optionality, and they want self-hosting, which they don’t get in the Vercel stack.”

Maintaining that open positioning is key. Companies building on open-source infrastructure often face tension between keeping their technology neutral and building a business. Atai emphasized that AG-UI is fully open, and CopilotKit’s commercial product hardens the open-source stack for enterprises without replacing it. “Our strategy is to be the default choice in the ecosystem, and then monetize the top enterprises,” added Uli, head of growth. “The open source must be the best out there, so 95% of users can build and get started without paying anyone.”

With around 25 employees, CopilotKit plans to use the new funding to expand its team.