assistant-ui logo/Docs

Generative UI

Render AI tool calls as interactive React components — charts, forms, maps, and custom widgets. Build generative UI patterns in chat with assistant-ui.

Create custom UI components for AI tool calls, providing visual feedback and interactive experiences when tools are executed.

Overview

Tool UIs in assistant-ui allow you to create custom interfaces that appear when AI tools are called. These generative UI components enhance the user experience by:

  • Visualizing tool execution with loading states and progress indicators
  • Displaying results in rich, formatted layouts
  • Enabling user interaction through forms and controls
  • Providing error feedback with helpful recovery options

This guide demonstrates building tool UIs with the Vercel AI SDK.

For composing UI from a JSON spec and component allowlist (display-only layouts), see the Generative UI primitive guide.

Creating Tool UIs

There are two main approaches to creating tool UIs in assistant-ui:

1. Client-Defined Tools (makeAssistantTool)

If you're creating tools on the client side, use makeAssistantTool to register them with the assistant context. Then create a UI component with makeAssistantToolUI. This component-based API coexists with the Tools() toolkit API; pick whichever fits your codebase better.

import { makeAssistantTool, tool } from "@assistant-ui/react";
import { z } from "zod";

// Define the tool
const weatherTool = tool({
  description: "Get current weather for a location",
  parameters: z.object({
    location: z.string(),
    unit: z.enum(["celsius", "fahrenheit"]),
  }),
  execute: async ({ location, unit }) => {
    const weather = await fetchWeatherAPI(location, unit);
    return weather;
  },
});

// Register the tool
const WeatherTool = makeAssistantTool({
  ...weatherTool,
  toolName: "getWeather",
});

// Create the UI
const WeatherToolUI = makeAssistantToolUI<
  { location: string; unit: "celsius" | "fahrenheit" },
  { temperature: number; description: string }
>({
  toolName: "getWeather",
  render: ({ args, result, status }) => {
    if (status.type === "running") {
      return <div>Checking weather in {args.location}...</div>;
    }

    return (
      <div className="weather-card">
        <h3>{args.location}</h3>
        <p>
          {result.temperature}°{args.unit === "celsius" ? "C" : "F"}
        </p>
        <p>{result.description}</p>
      </div>
    );
  },
});

Tools defined with makeAssistantTool can be passed to your backend using the frontendTools utility

Learn more about creating tools in the Tools Guide.

2. UI-Only for Existing Tools (makeAssistantToolUI)

If your tool is defined elsewhere (e.g., in your backend API, MCP server, or LangGraph), use makeAssistantToolUI to create just the UI component:

import { makeAssistantToolUI } from "@assistant-ui/react";

const WeatherToolUI = makeAssistantToolUI<
  { location: string; unit: "celsius" | "fahrenheit" },
  { temperature: number; description: string }
>({
  toolName: "getWeather", // Must match the backend tool name
  render: ({ args, result, status }) => {
    // UI rendering logic only
  },
});

Quick Start Example

This example shows how to implement the UI-only approach using makeAssistantToolUI:

Create a Tool UI Component

import { makeAssistantToolUI } from "@assistant-ui/react";
import { z } from "zod";

type WeatherArgs = {
  location: string;
  unit: "celsius" | "fahrenheit";
};

type WeatherResult = {
  temperature: number;
  description: string;
  humidity: number;
  windSpeed: number;
};

const WeatherToolUI = makeAssistantToolUI<WeatherArgs, WeatherResult>({
  toolName: "getWeather",
  render: ({ args, status, result }) => {
    if (status.type === "running") {
      return (
        <div className="flex items-center gap-2">
          <Spinner />
          <span>Checking weather in {args.location}...</span>
        </div>
      );
    }

    if (status.type === "incomplete" && status.reason === "error") {
      return (
        <div className="text-red-500">
          Failed to get weather for {args.location}
        </div>
      );
    }

    return (
      <div className="weather-card rounded-lg bg-blue-50 p-4">
        <h3 className="text-lg font-bold">{args.location}</h3>
        <div className="mt-2 grid grid-cols-2 gap-4">
          <div>
            <p className="text-2xl">
              {result.temperature}°{args.unit === "celsius" ? "C" : "F"}
            </p>
            <p className="text-gray-600">{result.description}</p>
          </div>
          <div className="text-sm">
            <p>Humidity: {result.humidity}%</p>
            <p>Wind: {result.windSpeed} km/h</p>
          </div>
        </div>
      </div>
    );
  },
});

Register the Tool UI

Place the component inside your AssistantRuntimeProvider:

function App() {
  return (
    <AssistantRuntimeProvider runtime={runtime}>
      <Thread />
      <WeatherToolUI />
    </AssistantRuntimeProvider>
  );
}

Define the Backend Tool (Vercel AI SDK)

When using the Vercel AI SDK, define the corresponding tool in your API route:

/app/api/chat/route.ts
import { streamText, tool, zodSchema } from "ai";
import { z } from "zod";

export async function POST(req: Request) {
  const { messages } = await req.json();

  const result = streamText({
    model: openai("gpt-5.4-nano"),
    messages: await convertToModelMessages(messages),
    tools: {
      getWeather: tool({
        description: "Get current weather for a location",
        inputSchema: zodSchema(
          z.object({
            location: z.string(),
            unit: z.enum(["celsius", "fahrenheit"]),
          }),
        ),
        execute: async ({ location, unit }) => {
          const weather = await fetchWeatherAPI(location);
          return {
            temperature: weather.temp,
            description: weather.condition,
            humidity: weather.humidity,
            windSpeed: weather.wind,
          };
        },
      }),
    },
  });

  return result.toUIMessageStreamResponse();
}

Tool UI Patterns

Component Pattern

Create standalone tool UI components:

export const WebSearchToolUI = makeAssistantToolUI<
  { query: string },
  { results: SearchResult[] }
>({
  toolName: "webSearch",
  render: ({ args, status, result }) => {
    return (
      <div className="search-container">
        <div className="mb-3 flex items-center gap-2">
          <SearchIcon />
          <span>Search results for: "{args.query}"</span>
        </div>

        {status.type === "running" && <LoadingSpinner />}

        {result && (
          <div className="space-y-2">
            {result.results.map((item, index) => (
              <div key={index} className="rounded border p-3">
                <a href={item.url} className="font-medium text-blue-600">
                  {item.title}
                </a>
                <p className="text-sm text-gray-600">{item.snippet}</p>
              </div>
            ))}
          </div>
        )}
      </div>
    );
  },
});

Hook Pattern

Use hooks for dynamic tool UI registration:

Use the useAssistantToolUI hook directly in your component for dynamic tool UI registration. This allows access to local component state and props when rendering the tool UI.

import { useAssistantToolUI } from "@assistant-ui/react";

function DynamicToolUI() {
  const [theme, setTheme] = useState("light");

  useAssistantToolUI({
    toolName: "analyzeData",
    render: ({ args, result, status }) => {
      // Hook allows access to component state
      return (
        <DataVisualization
          data={result}
          theme={theme}
          loading={status.type === "running"}
        />
      );
    },
  });

  return null;
}

Inline Pattern

For tools that need access to parent component props:

Why useInlineRender? By default, a tool UI's render function is static. Use useInlineRender when your UI needs access to dynamic component props (for example, to pass in an id or other contextual data).

import { useAssistantToolUI, useInlineRender } from "@assistant-ui/react";

function ProductPage({ productId, productName }) {
  useAssistantToolUI({
    toolName: "checkInventory",
    render: useInlineRender(({ args, result }) => {
      // Access parent component props
      return (
        <div className="inventory-status">
          <h4>{productName} Inventory</h4>
          <p>
            Stock for {productId}: {result.quantity} units
          </p>
          <p>Location: {result.warehouse}</p>
        </div>
      );
    }),
  });

  return <div>Product details...</div>;
}

Interactive Tool UIs

User Input Collection

Create tools that collect user input during execution:

Pro tip: Call addResult(...) exactly once to complete the tool call. After it's invoked, the assistant will resume the conversation with your provided data.

const DatePickerToolUI = makeAssistantToolUI<
  { prompt: string },
  { date: string }
>({
  toolName: "selectDate",
  render: ({ args, result, addResult }) => {
    if (result) {
      return (
        <div className="rounded bg-green-50 p-3">
          ✅ Selected date: {new Date(result.date).toLocaleDateString()}
        </div>
      );
    }

    return (
      <div className="rounded border p-4">
        <p className="mb-3">{args.prompt}</p>
        <DatePicker
          onChange={(date) => {
            addResult({ date: date.toISOString() });
          }}
        />
      </div>
    );
  },
});

Multi-Step Interactions

Build complex workflows with human-in-the-loop patterns for multi-step user interactions:

const DeleteProjectTool = makeAssistantTool({
  toolName: "deleteProject",
  parameters: z.object({
    projectId: z.string(),
  }),
  execute: async ({ projectId }, { human }) => {
    const response = await human({ action, details });
    if (!response.approved) throw new Error("Project deletion cancelled");

    await deleteProject(projectId);
    return { success: true };
  },
});

const ApprovalTool = makeAssistantTool({
  ...tool({
    description: "Request user approval for an action",
    parameters: z.object({
      action: z.string(),
      details: z.any(),
    }),
    execute: async ({ action, details }, { human }) => {
      // Request approval from user
      const response = await human({ action, details });

      return {
        approved: response.approved,
        reason: response.reason,
      };
    },
  }),
  toolName: "requestApproval",
  render: ({ args, result, interrupt, resume }) => {
    const [reason, setReason] = useState("");

    // Show result after approval/rejection
    if (result) {
      return (
        <div className={result.approved ? "text-green-600" : "text-red-600"}>
          {result.approved ? "✅ Approved" : `❌ Rejected: ${result.reason}`}
        </div>
      );
    }

    // Show approval UI when waiting for user input
    if (interrupt) {
      return (
        <div className="rounded border-2 border-yellow-400 p-4">
          <h4 className="font-bold">Approval Required</h4>
          <p className="my-2">{interrupt.payload.action}</p>
          <pre className="rounded bg-gray-100 p-2 text-sm">
            {JSON.stringify(interrupt.payload.details, null, 2)}
          </pre>

          <div className="mt-4 flex gap-2">
            <button
              onClick={() => resume({ approved: true })}
              className="rounded bg-green-500 px-4 py-2 text-white"
            >
              Approve
            </button>
            <button
              onClick={() => resume({ approved: false, reason })}
              className="rounded bg-red-500 px-4 py-2 text-white"
            >
              Reject
            </button>
            <input
              type="text"
              placeholder="Rejection reason..."
              value={reason}
              onChange={(e) => setReason(e.target.value)}
              className="flex-1 rounded border px-2"
            />
          </div>
        </div>
      );
    }

    return <div>Processing...</div>;
  },
});

Use tool human input (human() / resume()) for workflows that need to pause tool execution and wait for user input. Use addResult() for "human tools" where the AI requests a tool call but the entire execution happens through user interaction.

Server-side approval gates

Some runtimes (notably AI SDK v6's needsApproval tools) pause on the server and emit an approval request that the client must acknowledge before the tool runs. assistant-ui surfaces this on the tool part as approval and exposes respondToApproval({ approved, reason? }) on the renderer:

import { makeAssistantToolUI } from "@assistant-ui/react";

const DeployToolUI = makeAssistantToolUI<{ target: string }, { ok: boolean }>({
  toolName: "deploy",
  render: ({ args, approval, respondToApproval, result }) => {
    if (approval?.approved === undefined) {
      return (
        <div>
          <p>Approve deploy to {args.target}?</p>
          <button onClick={() => respondToApproval({ approved: true })}>
            Approve
          </button>
          <button
            onClick={() =>
              respondToApproval({ approved: false, reason: "user denied" })
            }
          >
            Deny
          </button>
        </div>
      );
    }

    if (approval?.approved === false) {
      return <p>Denied{approval.reason ? `: ${approval.reason}` : ""}</p>;
    }

    if (result === undefined) return <p>Approved, running…</p>;
    return <p>Deployed</p>;
  },
});

approval.approved is a three-state signal:

  • undefined: gate is open, the renderer should ask the user. This is the only state in which respondToApproval is legal.
  • true: decision recorded as allow. The server is producing the result (or has produced one, available on result).
  • false: decision recorded as deny. output-denied from the runtime sets isError and exposes approval.reason.

approval.isAutomatic is true when the runtime granted the decision from a server-side policy rather than the user; render a "auto-approved" badge instead of buttons in that case.

For the wire-side setup (needsApproval, sendAutomaticallyWhen), see AI SDK v6 server-side tool approval.

Advanced Features

Tool Status Handling

The status prop provides detailed execution state:

render: ({ status, args }) => {
  switch (status.type) {
    case "running":
      return <LoadingState />;

    case "requires-action":
      return <UserInputRequired reason={status.reason} />;

    case "incomplete":
      if (status.reason === "cancelled") {
        return <div>Operation cancelled</div>;
      }
      if (status.reason === "error") {
        return <ErrorDisplay error={status.error} />;
      }
      return <div>Failed: {status.reason}</div>;

    case "complete":
      return <SuccessDisplay />;
  }
};

Deferred Rendering

This section applies when the model drives the component through a tool call (args arrive incrementally and you want to wait for the final shape). If your backend or orchestrator pushes the component instead, prefer Data-Part Generative UI with makeAssistantDataUI. Data parts arrive as terminal events, so the renderer only fires once with the final data, no deferred rendering needed.

Sometimes you want to capture a tool call's streaming arguments but only render the final UI once the call completes. This is useful when partial args would render misleading or jarring intermediate states (a chart that flashes through half-populated data), when the component is expensive to mount (heavy visualizations, embedded iframes, third-party widgets), or when the model controls whether the component appears at all.

Inline at the end of streaming

Return null from the tool UI's render until status.type === "complete". The streaming args still arrive in args as the model emits them, you just ignore them until the call is done:

const ChartToolUI = makeAssistantToolUI<
  { title: string; series: number[] },
  void
>({
  toolName: "renderChart",
  render: ({ args, status }) => {
    if (status.type !== "complete") return null;
    return <Chart title={args.title} data={args.series} />;
  },
});

The chart mounts once, with the final args, after streaming finishes. No re-renders during the stream.

The same render shape works inside the Tools() toolkit's render field, with useAssistantToolUI, and with MessagePrimitive.Parts's inline tools.by_name overrides. The deferred-rendering pattern applies regardless of how you registered the tool UI.

Below the message body

If the component should sit outside the message parts (for example, a card attached under the avatar block rather than inline with text), gate at the message level with AuiIf and read s.message.status:

import { MessagePrimitive, AuiIf, useAuiState } from "@assistant-ui/react";

function PostMessageCard() {
  const parts = useAuiState((s) => s.message.parts);
  const chartCall = parts.find(
    (p) => p.type === "tool-call" && p.toolName === "renderChart",
  );
  if (!chartCall) return null;
  return <Chart {...chartCall.args} />;
}

<MessagePrimitive.Root>
  <MessagePrimitive.Parts />

  <AuiIf
    condition={(s) =>
      s.message.role === "assistant" &&
      s.message.status?.type === "complete"
    }
  >
    <PostMessageCard />
  </AuiIf>
</MessagePrimitive.Root>;

The AuiIf predicate fires whenever the assistant state changes; children mount only when both checks pass. PostMessageCard then reads the captured tool-call part from s.message.parts and renders from its args.

For the opposite pattern (showing partial data as it streams in), see Field-Level Streaming State and Partial Results & Streaming below.

Field-Level Streaming State

Use useToolArgsStatus to react to per-field streaming state. The hook returns a propStatus map where each top-level key in the args object resolves from "streaming" to "complete" as the partial JSON arrives. Call it inside a tool-call message part context:

import { useToolArgsStatus } from "@assistant-ui/react";

const FormToolUI = makeAssistantToolUI<{ email: string; phone: string }, unknown>({
  toolName: "submitForm",
  render: ({ args }) => {
    const { propStatus } = useToolArgsStatus<{ email: string; phone: string }>();

    return (
      <form className="space-y-4">
        <div>
          <input
            type="email"
            value={args.email ?? ""}
            className={propStatus.email === "streaming" ? "loading" : ""}
            disabled
          />
        </div>

        <div>
          <input
            type="tel"
            value={args.phone ?? ""}
            className={propStatus.phone === "streaming" ? "loading" : ""}
            disabled
          />
        </div>
      </form>
    );
  },
});

Partial Results & Streaming

Display results as they stream in:

const AnalysisToolUI = makeAssistantToolUI<
  { data: string },
  { progress: number; insights: string[] }
>({
  toolName: "analyzeData",
  render: ({ result, status }) => {
    const progress = result?.progress || 0;
    const insights = result?.insights || [];

    return (
      <div className="analysis-container">
        {status.type === "running" && (
          <div className="mb-4">
            <div className="mb-1 flex justify-between">
              <span>Analyzing...</span>
              <span>{progress}%</span>
            </div>
            <div className="w-full rounded bg-gray-200">
              <div
                className="h-2 rounded bg-blue-500"
                style={{ width: `${progress}%` }}
              />
            </div>
          </div>
        )}

        <div className="space-y-2">
          {insights.map((insight, i) => (
            <div key={i} className="rounded bg-gray-50 p-2">
              {insight}
            </div>
          ))}
        </div>
      </div>
    );
  },
});

Custom Tool Fallback

For tools that have no dedicated UI, add the ToolFallback shadcn component to your project. See the ToolFallback install guide for setup instructions and the ToolGroup guide for grouping consecutive tool calls into a collapsible container.

Execution Context

Generative UI components have access to execution context through props:

type ToolCallMessagePartProps<TArgs, TResult> = {
  // Tool arguments
  args: TArgs;
  argsText: string; // JSON stringified args

  // Execution status
  status: ToolCallMessagePartStatus;
  isError?: boolean;

  // Tool result (may be partial during streaming)
  result?: TResult;

  // Tool metadata
  toolName: string;
  toolCallId: string;

  // Interactive callbacks
  addResult: (result: TResult | ToolResponse<TResult>) => void;
  resume: (payload: unknown) => void;

  // Interrupt state
  interrupt?: { type: "human"; payload: unknown }; // Payload from context.human()

  // Optional artifact data
  artifact?: unknown;
};

Human Input Handling

When a tool calls human() during execution, the payload becomes available in the render function as interrupt.payload:

const ConfirmationToolUI = makeAssistantToolUI<
  { action: string },
  { confirmed: boolean }
>({
  toolName: "confirmAction",
  render: ({ args, result, interrupt, resume }) => {
    // Tool is waiting for user input
    if (interrupt) {
      return (
        <div className="confirmation-dialog">
          <p>Confirm: {interrupt.payload.message}</p>
          <button onClick={() => resume(true)}>Yes</button>
          <button onClick={() => resume(false)}>No</button>
        </div>
      );
    }

    // Tool completed
    if (result) {
      return <div>Action {result.confirmed ? "confirmed" : "cancelled"}</div>;
    }

    return <div>Processing...</div>;
  },
});

Learn more about tool human input in the Tools Guide.

Best Practices

1. Handle All Status States

Always handle loading, error, and success states:

render: ({ status, result, args }) => {
  if (status.type === "running") return <Skeleton />;
  if (status.type === "incomplete") return <ErrorState />;
  if (!result) return null;
  return <ResultDisplay result={result} />;
};

2. Provide Visual Feedback

Use animations and transitions for better UX:

<div
  className={cn(
    "transition-all duration-300",
    status.type === "running" && "opacity-50",
    status.type === "complete" && "opacity-100",
  )}
>
  {/* Tool UI content */}
</div>

3. Make UIs Accessible

Ensure keyboard navigation and screen reader support:

<button
  onClick={() => addResult(value)}
  aria-label="Confirm selection"
  className="focus:outline-none focus:ring-2"
>
  Confirm
</button>

4. Optimize Performance

Use useInlineRender to prevent unnecessary re-renders:

useAssistantToolUI({
  toolName: "heavyComputation",
  render: useInlineRender(({ result }) => {
    // Expensive rendering logic
    return <ComplexVisualization data={result} />;
  }),
});

Generative UI components are only displayed in the chat interface. The actual tool execution happens on the backend. This separation allows you to create rich, interactive experiences while keeping sensitive logic secure on the server.

Per-Property Streaming Status

When rendering a tool UI, you can track which arguments have finished streaming using useToolArgsStatus. This must be used inside a tool-call message part context.

import { useToolArgsStatus } from "@assistant-ui/react";

const WeatherUI = makeAssistantToolUI({
  toolName: "weather",
  render: ({ args }) => {
    const { status, propStatus } = useToolArgsStatus<{
      location: string;
      unit: string;
    }>();

    return (
      <div>
        <span className={propStatus.location === "streaming" ? "animate-pulse" : ""}>
          {args.location ?? "..."}
        </span>
        {status === "complete" && <WeatherChart data={args} />}
      </div>
    );
  },
});

propStatus maps each key to "streaming" | "complete" once the key appears in the partial JSON. Keys not yet present in the stream are absent from propStatus.

Data-Part Generative UI

Alongside tool-call rendering, assistant-ui supports a second generative UI mechanism based on DataMessagePart. Instead of attaching UI to a tool invocation, the backend (or the LangGraph graph) emits named data events that are appended as { type: "data", name, data } parts on the parent assistant message.

When to choose which:

  • Tool UI: the model decides what to render by calling a tool whose args become the component's data. Register the renderer via the Tools() toolkit's render field (recommended), or standalone with makeAssistantToolUI / useAssistantToolUI when the tool itself is defined elsewhere (backend, MCP, LangGraph). Args stream incrementally, so you observe partial state via status / useToolArgsStatus and may need Deferred Rendering for components that should only mount with final data.
  • Data UI (makeAssistantDataUI): the backend or orchestrator decides what to render and pushes a named data event onto the assistant message. Data parts arrive as terminal events with no streaming partials, so the renderer naturally fires once with the final data.

If you want a component to appear only after the message is complete and you control the backend, Data UI is usually the more direct fit; reach for Tool UI's deferred pattern when the model itself must drive the choice.

Use makeAssistantDataUI to register a renderer for a named data part:

import { makeAssistantDataUI } from "@assistant-ui/react";

type ChartProps = { series: number[]; title: string };

export const ChartUI = makeAssistantDataUI<ChartProps>({
  name: "chart",
  render: ({ data }) => (
    <div>
      <h3>{data.title}</h3>
      <Chart series={data.series} />
    </div>
  ),
});

Mount <ChartUI /> once inside the AssistantRuntimeProvider tree; it renders nothing itself and only registers the renderer.

For LangGraph-specific patterns (emitting UI from a Python/TypeScript graph node via push_ui_message / typedUi, dynamic loading with LoadExternalComponent, and the useLangGraphUIMessages escape hatch), see LangGraph Generative UI.

A fallback renderer for unmatched data parts is available internally but setFallbackDataUI is not yet a public API.