OpenAI ChatKit

by OpenAI

Embeddable, customizable chat UI components for agent experiences. See OpenAI AgentKit and OpenAI Agent Builder.

TL;DR

ChatKit provides pre-built chat components and developer integrations to embed agent-driven conversational experiences into web and mobile applications; it pairs with OpenAI AgentKit flows and can be exported from OpenAI Agent Builder.

Key features

• Theming & customization: style, layout, and behavior controls to match app branding.
• Streaming & partial responses: support for streamed outputs for better UX in long responses.
• Integration: direct export from OpenAI Agent Builder and SDK-compatible components for Agents SDK (Python/TypeScript).
• Tool integration: built-in support for connector-driven tool calls, file/document previews, and interactive approvals.
• Accessibility: keyboard navigation and ARIA-friendly components (implementation details depend on version).

Practical tips

• Use ChatKit export after finalizing flow in OpenAI Agent Builder to keep UI consistent with backend behavior.
• Implement local caching and rate-limiting for production to control costs when streaming long traces.

Who should use it

• Frontend teams building conversational interfaces backed by OpenAI AgentKit workflows.

Limitations

• May be opinionated in styling and behavior; teams with bespoke frontend needs might prefer custom UI built on Agents SDK.

Sources

• OpenAI announcement: “Introducing AgentKit” — https://openai.com/index/introducing-agentkit/