components（组件）

Tambo Components

Two component types: generative (AI creates on-demand) and interactable (pre-placed, AI updates).

Quick Start

// Generative: AI creates when needed
const components: TamboComponent[] = [
  {
    name: "WeatherCard",
    component: WeatherCard,
    description: "Shows weather. Use when user asks about weather.",
    propsSchema: z.object({ city: z.string(), temp: z.number() }),
  },
];

<TamboProvider components={components}>
  <App />
</TamboProvider>;

Generative Components

AI dynamically selects and renders these in response to user messages.

import { TamboProvider, TamboComponent } from "@tambo-ai/react";
import { z } from "zod";

const WeatherCardSchema = z.object({
  city: z.string().describe("City name"),
  temperature: z.number().describe("Temperature in Celsius"),
  condition: z.string().describe("Weather condition"),
});

const components: TamboComponent[] = [
  {
    name: "WeatherCard",
    component: WeatherCard,
    description:
      "Displays weather for a city. Use when user asks about weather.",
    propsSchema: WeatherCardSchema,
  },
];

<TamboProvider apiKey={apiKey} components={components}>
  <App />
</TamboProvider>;

Rendering Generative Components

Use ComponentRenderer to render AI-generated components in your message list:

import { ComponentRenderer } from "@tambo-ai/react";

function Message({
  message,
  threadId,
}: {
  message: TamboThreadMessage;
  threadId: string;
}) {
  return (
    <div>
      {message.content.map((block) => {
        switch (block.type) {
          case "text":
            return <p key={`${message.id}:text`}>{block.text}</p>;
          case "component":
            return (
              <ComponentRenderer
                key={block.id}
                content={block}
                threadId={threadId}
                messageId={message.id}
              />
            );
          default:
            return null;
        }
      })}
    </div>
  );
}

Generative Key Points

• propsSchema: Zod object with .describe() on each field
• description: Tells AI when to use the component
• Streaming: Props start undefined, make them optional or handle gracefully
• Use z.infer<typeof Schema> for TypeScript props type

Interactable Components

Pre-place in your UI; AI can observe and update props via natural language.

import { withTamboInteractable } from "@tambo-ai/react";
import { z } from "zod";

const NoteSchema = z.object({
  title: z.string().describe("Note title"),
  content: z.string().describe("Note content"),
  color: z.enum(["white", "yellow", "blue"]).optional(),
});

function Note({ title, content, color = "white" }: Props) {
  return (
    <div style={{ backgroundColor: color }}>
      <h3>{title}</h3>
      <p>{content}</p>
    </div>
  );
}

export const InteractableNote = withTamboInteractable(Note, {
  componentName: "Note",
  description: "A note with editable title, content, and color",
  propsSchema: NoteSchema,
});

Interactable How It Works

1. Auto-registration: Component registers when mounted
2. Context sending: Current props automatically visible to AI
3. Tool registration: Update tools registered automatically
4. Bidirectional: User edits and AI updates both work

When to Use Each