Tambo — Generative UI SDK for React

Build agents that speak your UI. Register your components — Tambo handles streaming, state management, and MCP.

Links

• GitHub: https://github.com/tambo-ai/tambo
• Website: https://tambo.co
• Docs: https://docs.tambo.co
• npm: @tambo-ai/react
• Discord: https://discord.gg/dJNvPEHth6
• Twitter: @tambo_ai
• Announcement: Introducing Tambo: Generative UI for React

What Is It

Tambo is an open-source React toolkit for building agents that render UI — sometimes called "generative UI." Instead of an AI returning text, it picks the right React component and streams props directly to it.

"Show me sales by region" renders your <Chart>. "Add a task" updates your <TaskBoard>."

You register components with Zod schemas. Those schemas become LLM tool definitions — the agent calls them like functions, and Tambo renders the result. The UI adapts to what the user asks for rather than being fixed.

What Problem Does It Solve

Most chat AI UIs return text that describes actions or data. Tambo lets the AI render the right interface — a chart, a form, a task board — directly in response to user intent. The UI becomes an output of the agent, not just a wrapper around it.

Key Features

Two Component Types

Generative components — render once in response to a message (charts, summaries, visualizations):

const components = [{
  name: "Graph",
  description: "Displays data as charts",
  component: Graph,
  propsSchema: z.object({
    data: z.array(z.object({ name: z.string(), value: z.number() })),
    type: z.enum(["line", "bar", "pie"]),
  }),
}];

Interactable components — persist and update as users refine requests (shopping carts, spreadsheets, task boards):

const InteractableNote = withInteractable(Note, {
  componentName: "Note",
  description: "A note supporting title, content, and color modifications",
  propsSchema: z.object({ title: z.string(), content: z.string(), color: z.enum(["white", "yellow", "blue", "green"]).optional() }),
});

Fullstack Solution

• Agent included — Tambo runs the LLM conversation loop; bring your own API key (OpenAI, Anthropic, Gemini, Mistral, or any OpenAI-compatible provider)
• Streaming infrastructure — props stream to components as the LLM generates them; cancellation, error recovery, and reconnection handled automatically
• Tambo Cloud or self-host — Cloud is a hosted backend; self-hosted runs via Docker

MCP Integration

Full MCP protocol support: tools, prompts, elicitations, and sampling. Connect to Linear, Slack, databases, or custom MCP servers:

<TamboProvider mcpServers={[{ name: "filesystem", url: "http://localhost:8261/mcp", transport: MCPTransport.HTTP }]}>

Other Features

• Local tools — functions that run in the browser (DOM manipulation, authenticated fetches, React state access)
• Context helpers — pass user state, app settings, current page to AI for better responses
• Suggestions — generates clickable prompts based on what users are doing
• Pre-built UI library — ui.tambo.co with agent and generative UI primitives

Quick Start

npm create tambo-app my-tambo-app
cd my-tambo-app
npm run dev

How Tambo Compares

Why It Matters

Tambo represents a shift in how AI integrates with application UIs. Instead of the AI being a chat sidebar that describes things, it becomes a first-class citizen that renders the interface. The application UI becomes dynamic — shaped by user intent rather than hardcoded routes and views.

This is part of a broader trend toward "agent-native" application design where the AI doesn't just assist — it drives UI state.

License

MIT (SDK), Apache-2.0 (API backend)

Related

• moc-planning-work — agent-driven workflows that could use generative UI for output
• temporal-io-ai — orchestration layer that could drive Tambo agents
• openspec.dev — spec-first approach compatible with Tambo's schema-driven component registration
• exe.dev — AI coding environment where tools like Tambo get built

---

Added: February 18, 2026