Building AI Agents with OpenAI AgentKit

OpenAI AgentKit represents a complete paradigm shift in how developers approach AI agent development. Announced at DevDay 2025, AgentKit is a comprehensive toolkit designed to simplify the entire lifecycle of AI agent creation, from initial development to production deployment and optimization.

The Four Core Components

AgentKit consists of four integrated components that work seamlessly together:

1. Agent Builder

2. ChatKit

Agent Builder provides a visual canvas for composing agent logic using drag-and-drop nodes. Developers can connect tools, configure custom guardrails, and design complex workflows without writing extensive boilerplate code. CEO Sam Altman described it as "like Canva for building agents."

ChatKit offers embeddable chat interfaces that developers can integrate into their own applications. This component handles the complexities of chat UI/UX, allowing developers to focus on agent logic rather than interface design.

3. Evals for Agents

4. Connector Registry

The evaluation toolkit includes step-by-step trace grading, datasets for assessing individual agent components, automated prompt optimization, and the ability to run evaluations on external models directly from the OpenAI platform.

The Connector Registry enables developers to securely connect agents to internal tools and third-party systems through an admin control panel while maintaining security and granular access control.

Real-World Impact

Early adopters have already seen remarkable results. Klarna built a support agent that now handles two-thirds of all customer tickets, while Clay achieved 10x growth by implementing a sales agent powered by AgentKit.

Getting Started

AgentKit launched in beta on October 6, 2025, and is available today for OpenAI API customers. Developers can access comprehensive documentation and start building immediately through the OpenAI platform.