AI Agents & UI Skills

AI Agents & UI Skills

Status: AI SDK 6 Beta Package Manager: pnpm Key Version: @ai-sdk/core, @ai-sdk/react (for UI) Official Docs:

• Agents Overview
• Building Agents
• Workflow Patterns
• AI SDK UI
• Chatbot Guide
• Chatbot Tool Usage
• Generative User Interfaces

---

Table of Contents

1. Installation & Setup
2. Understanding Agents
3. Building ToolLoopAgent
4. Agent Configuration Options
5. System Instructions & Behavior
6. Workflow Patterns
7. AI SDK UI - useChat Hook
8. Building Chatbot Applications
9. Tool Calling in UI
10. Generative User Interfaces
11. Advanced Patterns
12. Best Practices

---

Installation & Setup

Install Dependencies

# Core packages
pnpm add ai @ai-sdk/core @ai-sdk/anthropic

# For UI (React)
pnpm add @ai-sdk/react

# Optional: specific providers
pnpm add @ai-sdk/openai @ai-sdk/google

TypeScript Configuration

Ensure your tsconfig.json has proper settings:

{
  "compilerOptions": {
    "target": "ES2020",
    "lib": ["ES2020", "DOM"],
    "moduleResolution": "bundler",
    "strict": true
  }
}

---

Understanding Agents

What Are Agents?

Agents are LLMs that use tools in a loop to accomplish tasks. Three core components work together:

1. LLMs - Process input, decide next action
2. Tools - Extend capabilities (APIs, databases, files)
3. Loop - Orchestrates execution through context management and stopping conditions

ToolLoopAgent Class

The ToolLoopAgent class is the recommended approach because it:

• Reduces boilerplate - Manages loops and message arrays automatically
• Improves reusability - Define once, use throughout application
• Simplifies maintenance - Single place to update configuration
• Provides type safety - Full TypeScript support for tools and outputs

vs. Core Functions

For most use cases, use ToolLoopAgent. Use core functions (generateText, streamText) when you need explicit control for complex structured workflows.

---

Building ToolLoopAgent

Basic Agent

import { ToolLoopAgent, tool } from 'ai';
import { z } from 'zod';

const weatherAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  tools: {
    weather: tool({
      description: 'Get weather in a location (Fahrenheit)',
      inputSchema: z.object({
        location: z.string().describe('The location'),
      }),
      execute: async ({ location }) => ({
        location,
        temperature: 72 + Math.floor(Math.random() * 21) - 10,
      }),
    }),
    convertFahrenheitToCelsius: tool({
      description: 'Convert Fahrenheit to Celsius',
      inputSchema: z.object({
        temperature: z.number(),
      }),
      execute: async ({ temperature }) => ({
        celsius: Math.round((temperature - 32) * (5 / 9)),
      }),
    }),
  },
});

// Use the agent
const result = await weatherAgent.generate({
  prompt: 'What is the weather in San Francisco in celsius?',
});

console.log(result.text); // Agent's final answer
console.log(result.steps); // Steps taken by agent

Multi-Tool Execution

The agent automatically:

1. Calls weather tool to get temperature in Fahrenheit
2. Calls convertFahrenheitToCelsius to convert it
3. Generates final text response with result

---

Agent Configuration Options

Model and System Instructions

const agent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: 'You are an expert data analyst. Provide clear insights.',
});

Tools

const codeAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  tools: {
    runCode: tool({
      description: 'Execute Python code',
      inputSchema: z.object({
        code: z.string(),
      }),
      execute: async ({ code }) => {
        // Execute code
        return { output: 'Result' };
      },
    }),
  },
});

Loop Control (stopWhen)

By default, agents run for 20 steps (stopWhen: stepCountIs(20)). Each step is one generation (text or tool call).

import { ToolLoopAgent, stepCountIs } from 'ai';

const agent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  stopWhen: stepCountIs(20), // Allow up to 20 steps
});

// Combine multiple stop conditions
const agent2 = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  stopWhen: [
    stepCountIs(20),
    yourCustomCondition(), // Custom logic
  ],
});

The loop stops when:

• Finish reasoning (non-tool-call) is returned
• Tool without execute function is invoked
• Tool call needs approval
• Stop condition is met

Tool Choice

Control how agent uses tools:

const agent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  tools: { /* ... */ },
  toolChoice: 'required', // Force tool use
  // or 'none' to disable tools
  // or 'auto' (default) to let model decide
});

// Force specific tool
const agent2 = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  tools: { weather: weatherTool, attractions: attractionsTool },
  toolChoice: {
    type: 'tool',
    toolName: 'weather', // Force weather tool
  },
});

Structured Output

Define structured output schemas:

import { ToolLoopAgent, Output, stepCountIs } from 'ai';

const analysisAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  output: Output.object({
    schema: z.object({
      sentiment: z.enum(['positive', 'neutral', 'negative']),
      summary: z.string(),
      keyPoints: z.array(z.string()),
    }),
  }),
  stopWhen: stepCountIs(10),
});

const { output } = await analysisAgent.generate({
  prompt: 'Analyze customer feedback',
});

---

System Instructions & Behavior

Basic System Instructions

const agent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: 'You are an expert software engineer.',
});

Detailed Behavioral Instructions

const codeReviewAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: `You are a senior software engineer conducting code reviews.

Your approach:
- Focus on security vulnerabilities first
- Identify performance bottlenecks
- Suggest improvements for readability and maintainability
- Be constructive and educational in your feedback
- Always explain why something is an issue and how to fix it`,
});

Constrain Agent Behavior

const supportAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: `You are a customer support specialist.

Rules:
- Never make promises about refunds without checking policy
- Always be empathetic and professional
- If you don't know something, say so and offer to escalate
- Keep responses concise and actionable
- Never share internal company information`,
  tools: {
    checkOrderStatus,
    lookupPolicy,
    createTicket,
  },
});

Tool Usage Instructions

const researchAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: `You are a research assistant with access to search and document tools.

When researching:
1. Always start with a broad search to understand the topic
2. Use document analysis for detailed information
3. Cross-reference multiple sources before drawing conclusions
4. Cite your sources when presenting information
5. If information conflicts, present both viewpoints`,
  tools: {
    webSearch,
    analyzeDocument,
    extractQuotes,
  },
});

Format and Style Instructions

const technicalWriterAgent = new ToolLoopAgent({
  model: 'anthropic/claude-sonnet-4.5',
  instructions: `You are a technical documentation writer.

Writing style:
- Use clear, simple language
- Avoid jargon unless necessary
- Structure information with headers and bullet points
- Include code examples where relevant
- Write in second person ("you" instead of "the user")

Always format responses in Markdown.`,
});

---

Workflow Patterns

Combine patterns thoughtfully to build reliable solutions for complex problems.

1. Sequential Processing (Chains)

Execute steps in predefined order. Each step's output becomes next step's input.

Use case: Content generation pipelines, data transformation processes

import { generateText, generateObject } from 'ai';
import { z } from 'zod';

async function generateMarketingCopy(input: string) {
  const model = 'anthropic/claude-sonnet-4.5';

  // Step 1: Generate marketing copy
  const { text: copy } = await generateText({
    model,
    prompt: `Write persuasive marketing copy for: ${input}. Focus on benefits and emotional appeal.`,
  });

  // Step 2: Quality check
  const { object: qualityMetrics } = await generateObject({
    model,
    schema: z.object({
      hasCallToAction: z.boolean(),
      emotionalAppeal: z.number().min(1).max(10),
      clarity: z.number().min(1).max(10),
    }),
    prompt: `Evaluate this marketing copy:
1. Presence of call to action (true/false)
2. Emotional appeal (1-10)
3. Clarity (1-10)

Copy to evaluate: ${copy}`,
  });

  // Step 3: Regenerate if needed
  if (
    !qualityMetrics.hasCallToAction ||
    qualityMetrics.emotionalAppeal < 7 ||
    qualityMetrics.clarity < 7
  ) {
    const { text: improvedCopy } = await generateText({
      model,
      prompt: `Rewrite this marketing copy with:
${!qualityMetrics.hasCallToAction ? '- A clear call to action' : ''}
${qualityMetrics.emotionalAppeal < 7 ? '- Stronger emotional appeal' : ''}
${qualityMetrics.clarity < 7 ? '- Improved clarity and directness' : ''}

Original copy: ${copy}`,
    });
    return { copy: improvedCopy, qualityMetrics };
  }

  return { copy, qualityMetrics };
}

2. Routing

Let the model decide which path to take based on context and intermediate results.

Use case: Handling varied inputs that require different processing approaches

import { generateObject, generateText } from 'ai';
import { z } from 'zod';

async function handleCustomerQuery(query: string) {
  const model = 'anthropic/claude-sonnet-4.5';

  // Step 1: Classify the query
  const { object: classification } = await generateObject({
    model,
    schema: z.object({
      reasoning: z.string(),
      type: z.enum(['general', 'refund', 'technical']),
      complexity: z.enum(['simple', 'complex']),
    }),
    prompt: `Classify this customer query: ${query}

Determine:
1. Query type (general, refund, or technical)
2. Complexity (simple or complex)
3. Brief reasoning`,
  });

  // Step 2: Route based on classification
  const { text: response } = await generateText({
    model:
      classification.complexity === 'simple'
        ? 'openai/gpt-4o-mini'
        : 'openai/o4-mini',
    system: {
      general: 'You are an expert customer service agent.',
      refund: 'You are a refund specialist. Follow company policy strictly.',
      technical: 'You are a technical support specialist.',
    }[classification.type],
    prompt: query,
  });

  return { response, classification };
}

3. Parallel Processing

Break tasks into independent subtasks that execute simultaneously.

Use case: Analyzing multiple documents, different aspects of input (like code review)

import { generateText, generateObject } from 'ai';
import { z } from 'zod';

async function parallelCodeReview(code: string) {
  const model = 'anthropic/claude-sonnet-4.5';

  // Run parallel reviews
  const [securityReview, performanceReview, maintainabilityReview] =
    await Promise.all([
      generateObject({
        model,
        system: 'You are an expert in code security.',
        schema: z.object({
          vulnerabilities: z.array(z.string()),
          riskLevel: z.enum(['low', 'medium', 'high']),
          suggestions: z.array(z.string()),
        }),
        prompt: `Review this code: ${code}`,
      }),
      generateObject({
        model,
        system: 'You are an expert in code performance.',
        schema: z.object({
          issues: z.array(z.string()),
          impact: z.enum(['low', 'medium', 'high']),
          optimizations: z.array(z.string()),
        }),
        prompt: `Review this code: ${code}`,
      }),
      generateObject({
        model,
        system: 'You are an expert in code quality.',
        schema: z.object({
          concerns: z.array(z.string()),
          qualityScore: z.number().min(1).max(10),
          recommendations: z.array(z.string()),
        }),
        prompt: `Review this code: ${code}`,
      }),
    ]);

  // Aggregate results
  const { text: summary } = await generateText({
    model,
    system: 'You are a technical lead summarizing multiple code reviews.',
    prompt: `Synthesize these code review results into a concise summary:
${JSON.stringify(
  [
    { ...securityReview.object, type: 'security' },
    { ...performanceReview.object, type: 'performance' },
    { ...maintainabilityReview.object, type: 'maintainability' },
  ],
  null,
  2,
)}`,
  });

  return { reviews: [securityReview.object, performanceReview.object, maintainabilityReview.object], summary };
}

4. Orchestrator-Worker

Primary model coordinates specialized workers. Each worker optimizes for specific subtask.

Use case: Complex tasks requiring different types of expertise

import { generateObject } from 'ai';
import { z } from 'zod';

async function implementFeature(featureRequest: string) {
  // Orchestrator: Plan the implementation
  const { object: implementationPlan } = await generateObject({
    model: 'anthropic/claude-sonnet-4.5',
    schema: z.object({
      files: z.array(
        z.object({
          purpose: z.string(),
          filePath: z.string(),
          changeType: z.enum(['create', 'modify', 'delete']),
        }),
      ),
      estimatedComplexity: z.enum(['low', 'medium', 'high']),
    }),
    system:
      'You are a senior software architect planning feature implementations.',
    prompt: `Analyze this feature request and create an implementation plan: ${featureRequest}`,
  });

  // Workers: Execute the planned changes
  const fileChanges = await Promise.all(
    implementationPlan.files.map(async (file) => {
      const workerSystemPrompt = {
        create:
          'You are an expert at implementing new files following best practices.',
        modify:
          'You are an expert at modifying existing code while maintaining consistency.',
        delete:
          'You are an expert at safely removing code while ensuring no breaking changes.',
      }[file.changeType];

      const { object: change } = await generateObject({
        model: 'anthropic/claude-sonnet-4.5',
        schema: z.object({
          explanation: z.string(),
          code: z.string(),
        }),
        system: workerSystemPrompt,
        prompt: `Implement changes for ${file.filePath} to support: ${file.purpose}`,
      });

      return {
        file,
        implementation: change,
      };
    }),
  );

  return {
    plan: implementationPlan,
    changes: fileChanges,
  };
}

5. Evaluator-Optimizer

Add quality control with dedicated evaluation steps that assess intermediate results.

Use case: Robust workflows with self-improvement and error recovery

import { generateText, generateObject } from 'ai';
import { z } from 'zod';

async function translateWithFeedback(text: string, targetLanguage: string) {
  let currentTranslation = '';
  let iterations = 0;
  const MAX_ITERATIONS = 3;

  // Initial translation
  const { text: translation } = await generateText({
    model: 'anthropic/claude-sonnet-4.5',
    system: 'You are an expert literary translator.',
    prompt: `Translate this text to ${targetLanguage}, preserving tone and cultural nuances: ${text}`,
  });

  currentTranslation = translation;

  // Evaluation-optimization loop
  while (iterations < MAX_ITERATIONS) {
    // Evaluate current translation
    const { object: evaluation } = await generateObject({
      model: 'anthropic/claude-sonnet-4.5',
      schema: z.object({
        qualityScore: z.number().min(1).max(10),
        preservesTone: z.boolean(),
        preservesNuance: z.boolean(),
        culturallyAccurate: z.boolean(),
        specificIssues: z.array(z.string()),
        improvementSuggestions: z.array(z.string()),
      }),
      system: 'You are an expert in evaluating literary translations.',
      prompt: `Evaluate this translation:
Original: ${text}
Translation: ${currentTranslation}

Consider: tone, nuance, cultural accuracy`,
    });

    // Check if quality meets threshold
    if (
      evaluation.qualityScore >= 8 &&
      evaluation.preservesTone &&
      evaluation.preservesNuance &&
      evaluation.culturallyAccurate
    ) {
      break;
    }

    // Generate improved translation based on feedback
    const { text: improvedTranslation } = await generateText({
      model: 'anthropic/claude-sonnet-4.5',
      system: 'You are an expert literary translator.',
      prompt: `Improve this translation based on feedback:
${evaluation.specificIssues.join('\n')}
${evaluation.improvementSuggestions.join('\n')}

Original: ${text}
Current Translation: ${currentTranslation}`,
    });

    currentTranslation = improvedTranslation;
    iterations++;
  }

  return {
    finalTranslation: currentTranslation,
    iterationsRequired: iterations,
  };
}

---

AI SDK UI - useChat Hook

The useChat hook makes it effortless to create a conversational UI for chatbot applications.

Features

• Message Streaming - Real-time streaming from AI provider
• Managed States - Input, messages, status, error states
• Seamless Integration - Easy integration into any design

Basic Example

Client:

'use client';

import { useChat } from '@ai-sdk/react';
import { DefaultChatTransport } from 'ai';
import { useState } from 'react';

export default function Page() {
  const { messages, sendMessage, status } = useChat({
    transport: new DefaultChatTransport({
      api: '/api/chat',
    }),
  });
  const [input, setInput] = useState('');

  return (
    <>
      {messages.map((message) => (
        <div key={message.id}>
          {message.role === 'user' ? 'User: ' : 'AI: '}
          {message.parts.map((part, index) =>
            part.type === 'text' ? <span key={index}>{part.text}</span> : null,
          )}
        </div>
      ))}

      <form
        onSubmit={(e) => {
          e.preventDefault();
          if (input.trim()) {
            sendMessage({ text: input });
            setInput('');
          }
        }}
      >
        <input
          value={input}
          onChange={(e) => setInput(e.target.value)}
          disabled={status !== 'ready'}
          placeholder="Say something..."
        />
        <button type="submit" disabled={status !== 'ready'}>
          Submit
        </button>
      </form>
    </>
  );
}

Server:

import { convertToModelMessages, streamText, UIMessage } from 'ai';

export const maxDuration = 30;

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

  const result = streamText({
    model: 'anthropic/claude-sonnet-4.5',
    system: 'You are a helpful assistant.',
    messages: await convertToModelMessages(messages),
  });

  return result.toUIMessageStreamResponse();
}

Message Parts

Messages use a parts property that contains message parts (text, tool invocation, tool result):

{messages.map((message) => (
  <div key={message.id}>
    {message.parts.map((part, index) => {
      if (part.type === 'text') {
        return <span key={index}>{part.text}</span>;
      }
      // Handle other part types
    })}
  </div>
))}

Status States

const { messages, sendMessage, status, stop } = useChat({
  transport: new DefaultChatTransport({
    api: '/api/chat',
  }),
});

// Status values:
// - 'submitted': Message sent, awaiting response start
// - 'streaming': Response actively streaming
// - 'ready': Full response received
// - 'error': Error occurred

{(status === 'submitted' || status === 'streaming') && (
  <div>
    {status === 'submitted' && <Spinner />}
    <button type="button" onClick={() => stop()}>
      Stop
    </button>
  </div>
)}

Error Handling

const { messages, sendMessage, error, reload } = useChat({
  transport: new DefaultChatTransport({
    api: '/api/chat',
  }),
});

{error && (
  <>
    <div>An error occurred.</div>
    <button type="button" onClick={() => reload()}>
      Retry
    </button>
  </>
)}

Modify Messages

const { messages, setMessages } = useChat();

const handleDelete = (id: string) => {
  setMessages(messages.filter((message) => message.id !== id));
};

{messages.map((message) => (
  <div key={message.id}>
    {/* message content */}
    <button onClick={() => handleDelete(message.id)}>Delete</button>
  </div>
))}

Cancellation and Regeneration

const { stop, status, regenerate } = useChat();

<button onClick={stop} disabled={!(status === 'streaming' || status === 'submitted')}>
  Stop
</button>

<button onClick={regenerate} disabled={!(status === 'ready' || status === 'error')}>
  Regenerate
</button>

Event Callbacks

import { UIMessage } from 'ai';

const { messages } = useChat({
  onFinish: ({ message, messages, isAbort, isDisconnect, isError }) => {
    // Use information to update other UI states
  },
  onError: (error) => {
    console.error('An error occurred:', error);
  },
  onData: (data) => {
    console.log('Received data part from server:', data);
  },
});

Request Configuration

import { useChat } from '@ai-sdk/react';
import { DefaultChatTransport } from 'ai';

const { messages, sendMessage } = useChat({
  transport: new DefaultChatTransport({
    api: '/api/custom-chat',
    headers: {
      Authorization: 'your_token',
    },
    body: {
      user_id: '123',
    },
  }),
});

// Dynamic configuration
const { messages: messages2 } = useChat({
  transport: new DefaultChatTransport({
    api: '/api/chat',
    headers: () => ({
      Authorization: `Bearer ${getAuthToken()}`,
      'X-User-ID': getCurrentUserId(),
    }),
    body: () => ({
      sessionId: getCurrentSessionId(),
      preferences: getUserPreferences(),
    }),
  }),
});

// Request-level configuration (recommended)
sendMessage(
  { text: input },
  {
    headers: {
      Authorization: 'Bearer token123',
      'X-Custom-Header': 'custom-value',
    },
    body: {
      temperature: 0.7,
      max_tokens: 100,
      user_id: '123',
    },
  },
);

---

Building Chatbot Applications

Full Chatbot Example with Status and Error Handling

'use client';

import { useChat } from '@ai-sdk/react';
import { DefaultChatTransport } from 'ai';
import { useState } from 'react';

export default function Chat() {
  const { messages, sendMessage, error, reload, status, stop } = useChat({
    transport: new DefaultChatTransport({
      api: '/api/chat',
    }),
  });
  const [input, setInput] = useState('');

  return (
    <div className="flex flex-col w-full max-w-md py-24 mx-auto stretch">
      {messages.map((m) => (
        <div key={m.id} className="whitespace-pre-wrap">
          {m.role === 'user' ? 'User: ' : 'AI: '}
          {m.parts.map((part, i) =>
            part.type === 'text' ? <span key={i}>{part.text}</span> : null,
          )}
        </div>
      ))}

      {status === 'error' && (
        <div className="text-red-500">
          <p>An error occurred. Please try again.</p>
          <button onClick={() => reload()}>Retry</button>
        </div>
      )}

      {(status === 'submitted' || status === 'streaming') && (
        <div>
          <button onClick={() => stop()}>Stop</button>
        </div>
      )}

      <form
        onSubmit={(e) => {
          e.preventDefault();
          if (input.trim() && status === 'ready') {
            sendMessage({ text: input });
            setInput('');
          }
        }}
      >
        <input
          value={input}
          onChange={(e) => setInput(e.target.value)}
          disabled={status !== 'ready'}
          placeholder="Send message..."
          className="w-full p-2 border rounded"
        />
        <button type="submit" disabled={status !== 'ready'}>
          Send
        </button>
      </form>
    </div>
  );
}

Message Metadata

Track token consumption and usage information:

import { LanguageModelUsage, UIMessage, streamText, convertToModelMessages } from 'ai';

// Create custom metadata type
type MyMetadata = {
  totalUsage: LanguageModelUsage;
};

export type MyUIMessage = UIMessage<MyMetadata>;

// Server
export async function POST(req: Request) {
  const { messages }: { messages: MyUIMessage[] } = await req.json();

  const result = streamText({
    model: 'anthropic/claude-sonnet-4.5',
    messages: await convertToModelMessages(messages),
  });

  return result.toUIMessageStreamResponse({
    originalMessages: messages,
    messageMetadata: ({ part }) => {
      if (part.type === 'finish') {
        return { totalUsage: part.totalUsage };
      }
    },
  });
}

// Client
export default function Chat() {
  const { messages } = useChat<MyUIMessage>({
    transport: new DefaultChatTransport({
      api: '/api/chat',
    }),
  });

  return (
    <div>
      {messages.map((m) => (
        <div key={m.id}>
          {m.role === 'user' ? 'User: ' : 'AI: '}
          {m.parts.map((part) => {
            if (part.type === 'text') {
              return part.text;
            }
          })}
          {m.metadata?.totalUsage && (
            <div>Total usage: {m.metadata?.totalUsage.totalTokens} tokens</div>
          )}
        </div>
      ))}
    </div>
  );
}

---

Tool Calling in UI

Three Types of Tools in Chatbot

1. Automatically executed server-side tools - Execute on server, use execute method
2. Automatically executed client-side tools - Execute in browser, handle with onToolCall
3. Tools requiring user interaction - Confirmation dialogs, displayed in UI

Tool Call Flow

1. User enters message
2. Message sent to API route
3. Language model generates tool calls during streamText
4. Tool calls forwarded to client
5. Server-side tools executed, results forwarded
6. Client-side tools handled with onToolCall callback
7. User-interaction tools displayed in UI
8. addToolOutput provides tool result
9. Chat configured to auto-submit when all results available

Full Tool Usage Example

Server:

import { convertToModelMessages, streamText, UIMessage } from 'ai';
import { z } from 'zod';

export const maxDuration = 30;

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

  const result = streamText({
    model: 'anthropic/claude-sonnet-4.5',
    messages: await convertToModelMessages(messages),
    tools: {
      // Server-side tool with execute
      getWeatherInformation: {
        description: 'Show the weather in a given city to the user',
        inputSchema: z.object({ city: z.string() }),
        execute: async ({ city }: { city: string }) => {
          const weatherOptions = ['sunny', 'cloudy', 'rainy', 'snowy', 'windy'];
          return weatherOptions[Math.floor(Math.random() * weatherOptions.length)];
        },
      },
      // Client-side tool for user interaction
      askForConfirmation: {
        description: 'Ask the user for confirmation.',
        inputSchema: z.object({
          message: z.string().describe('The message to ask for confirmation.'),
        }),
      },
      // Automatically executed client-side tool
      getLocation: {
        description: 'Get the user\'s location',
        inputSchema: z.object({
          query: z.string(),
        }),
      },
    },
  });

  return result.toUIMessageStreamResponse();
}

Client:

'use client';

import { useChat } from '@ai-sdk/react';
import { DefaultChatTransport, lastAssistantMessageIsCompleteWithToolCalls } from 'ai';
import { useState } from 'react';

export default function Chat() {
  const { messages, sendMessage, addToolOutput } = useChat({
    transport: new DefaultChatTransport({
      api: '/api/chat',
    }),

    sendAutomaticallyWhen: lastAssistantMessageIsCompleteWithToolCalls,

    // Run client-side tools automatically
    async onToolCall({ toolCall }) {
      // Always check if dynamic tool first
      if (toolCall.dynamic) {
        return;
      }

      if (toolCall.toolName === 'getLocation') {
        const cities = ['New York', 'Los Angeles', 'Chicago'];
        addToolOutput({
          tool: 'getLocation',
          toolCallId: toolCall.toolCallId,
          output: cities[Math.floor(Math.random() * cities.length)],
        });
      }
    },
  });
  const [input, setInput] = useState('');

  return (
    <>
      {messages?.map((message) => (
        <div key={message.id}>
          <strong>{`${message.role}: `}</strong>
          {message.parts.map((part) => {
            switch (part.type) {
              case 'text':
                return part.text;

              case 'tool-askForConfirmation': {
                const callId = part.toolCallId;
                switch (part.state) {
                  case 'input-available':
                    return (
                      <div key={callId}>
                        {part.input.message}
                        <button
                          onClick={() =>
                            addToolOutput({
                              tool: 'askForConfirmation',
                              toolCallId: callId,
                              output: 'Yes, confirmed.',
                            })
                          }
                        >
                          Yes
                        </button>
                        <button
                          onClick={() =>
                            addToolOutput({
                              tool: 'askForConfirmation',
                              toolCallId: callId,
                              output: 'No, denied',
                            })
                          }
                        >
                          No
                        </button>
                      </div>
                    );
                  case 'output-available':
                    return <div key={callId}>Result: {part.output}</div>;
                }
                break;
              }

              case 'tool-getWeatherInformation': {
                const callId = part.toolCallId;
                switch (part.state) {
                  case 'input-available':
                    return <div key={callId}>Getting weather for {part.input.city}...</div>;
                  case 'output-available':
                    return (
                      <div key={callId}>
                        Weather in {part.input.city}: {part.output}
                      </div>
                    );
                }
                break;
              }
            }
          })}
          <br />
        </div>
      ))}

      <form
        onSubmit={(e) => {
          e.preventDefault();
          if (input.trim()) {
            sendMessage({ text: input });
            setInput('');
          }
        }}
      >
        <input value={input} onChange={(e) => setInput(e.target.value)} />
      </form>
    </>
  );
}

Tool Execution Approval

Require user confirmation before server-side tool runs:

Server:

import { streamText, tool } from 'ai';
import { z } from 'zod';

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

  const result = streamText({
    model: 'anthropic/claude-sonnet-4.5',
    messages,
    tools: {
      getWeather: tool({
        description: 'Get the weather in a location',
        inputSchema: z.object({
          city: z.string(),
        }),
        needsApproval: true,
        execute: async ({ city }) => {
          const weather = await fetchWeather(city);
          return weather;
        },
      }),
    },
  });

  return result.toUIMessageStreamResponse();
}

Client:

'use client';

import { useChat } from '@ai-sdk/react';

export default function Chat() {
  const { messages, addToolApprovalResponse } = useChat();

  return (
    <>
      {messages.map((message) => (
        <div key={message.id}>
          {message.parts.map((part) => {
            if (part.type === 'tool-getWeather') {
              switch (part.state) {
                case 'approval-requested':
                  return (
                    <div key={part.toolCallId}>
                      <p>Get weather for {part.input.city}?</p>
                      <button
                        onClick={() =>
                          addToolApprovalResponse({
                            id: part.approval.id,
                            approved: true,
                          })
                        }
                      >
                        Approve
                      </button>
                      <button
                        onClick={() =>
                          addToolApprovalResponse({
                            id: part.approval.id,
                            approved: false,
                          })
                        }
                      >
                        Deny
                      </button>
                    </div>
                  );
                case 'output-available':
                  return (
                    <div key={part.toolCallId}>
                      Weather in {part.input.city}: {part.output}
                    </div>
                  );
              }
            }
          })}
        </div>
      ))}
    </>
  );
}

---

Generative User Interfaces

Generative UI allows LLMs to generate UI components, not just text.

Core Concept

1. You provide model with tools
2. Model decides when/how to use tools based on context
3. Tool executes and returns data
4. Data passed to React component for rendering

Basic Generative UI Example

Tools:

import { tool as createTool } from 'ai';
import { z } from 'zod';

export const weatherTool = createTool({
  description: 'Display the weather for a location',
  inputSchema: z.object({
    location: z.string().describe('The location'),
  }),
  execute: async function ({ location }) {
    await new Promise((resolve) => setTimeout(resolve, 2000));
    return { weather: 'Sunny', temperature: 75, location };
  },
});

export const tools = {
  displayWeather: weatherTool,
};

UI Component:

type WeatherProps = {
  temperature: number;
  weather: string;
  location: string;
};

export const Weather = ({ temperature, weather, location }: WeatherProps) => {
  return (
    <div>
      <h2>Current Weather for {location}</h2>
      <p>Condition: {weather}</p>
      <p>Temperature: {temperature}°C</p>
    </div>
  );
};

Chatbot:

'use client';

import { useChat } from '@ai-sdk/react';
import { useState } from 'react';
import { Weather } from '@/components/weather';

export default function Page() {
  const [input, setInput] = useState('');
  const { messages, sendMessage } = useChat();

  const handleSubmit = (e: React.FormEvent) => {
    e.preventDefault();
    sendMessage({ text: input });
    setInput('');
  };

  return (
    <div>
      {messages.map((message) => (
        <div key={message.id}>
          <div>{message.role === 'user' ? 'User: ' : 'AI: '}</div>
          <div>
            {message.parts.map((part, index) => {
              if (part.type === 'text') {
                return <span key={index}>{part.text}</span>;
              }

              if (part.type === 'tool-displayWeather') {
                switch (part.state) {
                  case 'input-available':
                    return <div key={index}>Loading weather...</div>;
                  case 'output-available':
                    return (
                      <div key={index}>
                        <Weather {...part.output} />
                      </div>
                    );
                  case 'output-error':
                    return <div key={index}>Error: {part.errorText}</div>;
                }
              }

              return null;
            })}
          </div>
        </div>
      ))}

      <form onSubmit={handleSubmit}>
        <input
          value={input}
          onChange={(e) => setInput(e.target.value)}
          placeholder="Type a message..."
        />
        <button type="submit">Send</button>
      </form>
    </div>
  );
}

Server:

import { streamText, convertToModelMessages, UIMessage, stepCountIs } from 'ai';
import { tools } from '@/ai/tools';

export async function POST(request: Request) {
  const { messages }: { messages: UIMessage[] } = await request.json();

  const result = streamText({
    model: 'anthropic/claude-sonnet-4.5',
    system: 'You are a friendly assistant!',
    messages: await convertToModelMessages(messages),
    stopWhen: stepCountIs(5),
    tools,
  });

  return result.toUIMessageStreamResponse();
}

Adding Multiple Tools

// Define more tools
export const stockTool = createTool({
  description: 'Get price for a stock',
  inputSchema: z.object({
    symbol: z.string(),
  }),
  execute: async ({ symbol }) => {
    await new Promise((resolve) => setTimeout(resolve, 2000));
    return { symbol, price: 150 };
  },
});

export const tools = {
  displayWeather: weatherTool,
  getStockPrice: stockTool,
};

Then render multiple components based on tool types:

{message.parts.map((part, index) => {
  if (part.type === 'tool-displayWeather') {
    return <Weather {...part.output} />;
  }
  if (part.type === 'tool-getStockPrice') {
    return <Stock {...part.output} />;
  }
})}

---

Advanced Patterns

Type Safety for Tools

import { InferUITool, InferUITools, UIMessage, UIDataTypes } from 'ai';

// Infer single tool type
type WeatherUITool = InferUITool<typeof weatherTool>;
// { input: { location: string }; output: string }

// Infer tool set types
type MyUITools = InferUITools<typeof tools>;

// Use in custom message type
type MyUIMessage = UIMessage<never, UIDataTypes, MyUITools>;

// Pass to useChat
const { messages } = useChat<MyUIMessage>();

Reasoning and Sources

Some models support reasoning tokens and sources:

Server:

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

  const result = streamText({
    model: 'deepseek/deepseek-r1',
    messages: await convertToModelMessages(messages),
  });

  return result.toUIMessageStreamResponse({
    sendReasoning: true, // Include reasoning tokens
    sendSources: true,   // Include sources
  });
}

Client:

{message.parts.map((part, index) => {
  if (part.type === 'text') {
    return <div key={index}>{part.text}</div>;
  }

  if (part.type === 'reasoning') {
    return <pre key={index}>{part.text}</pre>;
  }

  if (part.type === 'source-url') {
    return (
      <a key={`source-${part.id}`} href={part.url} target="_blank">
        {part.title ?? new URL(part.url).hostname}
      </a>
    );
  }

  if (part.type === 'source-document') {
    return <span key={`source-${part.id}`}>{part.title ?? `Document`}</span>;
  }
})}

File Attachments

'use client';

import { useChat } from '@ai-sdk/react';
import { useRef, useState } from 'react';

export default function Page() {
  const { messages, sendMessage, status } = useChat();
  const [input, setInput] = useState('');
  const [files, setFiles] = useState<FileList | undefined>();
  const fileInputRef = useRef<HTMLInputElement>(null);

  return (
    <div>
      {messages.map((message) => (
        <div key={message.id}>
          {message.parts.map((part, index) => {
            if (part.type === 'text') {
              return <span key={index}>{part.text}</span>;
            }

            if (part.type === 'file' && part.mediaType?.startsWith('image/')) {
              return <img key={index} src={part.url} alt={part.filename} />;
            }

            return null;
          })}
        </div>
      ))}

      <form
        onSubmit={(event) => {
          event.preventDefault();
          if (input.trim()) {
            sendMessage({
              text: input,
              files,
            });
            setInput('');
            setFiles(undefined);
            if (fileInputRef.current) {
              fileInputRef.current.value = '';
            }
          }
        }}
      >
        <input
          type="file"
          onChange={(event) => {
            if (event.target.files) {
              setFiles(event.target.files);
            }
          }}
          multiple
          ref={fileInputRef}
        />
        <input
          value={input}
          placeholder="Send message..."
          onChange={(e) => setInput(e.target.value)}
          disabled={status !== 'ready'}
        />
      </form>
    </div>
  );
}

---

Best Practices

Agents

1. Start with ToolLoopAgent - Use for most cases; only use core functions when you need explicit control
2. Clear system instructions - Define agent's role, expertise, and constraints clearly
3. Appropriate tool choice - Use 'required' when tools must be used, 'auto' for flexibility
4. Set reasonable step limits - Balance thorough execution with resource consumption
5. Tool descriptions matter - Provide clear, specific descriptions so model uses tools correctly

Workflows

1. Choose appropriate pattern - Use simplest pattern that meets requirements
2. Plan before implementation - Break down complex tasks into clear steps
3. Add error handling - Gracefully handle tool failures and edge cases
4. Monitor costs - Complex workflows mean more LLM calls
5. Test extensively - Non-deterministic nature requires thorough testing

Chat UI

1. Handle all status states - Show spinners, error states, and stop buttons
2. Use typed message parts - Leverage tool-${toolName} typed parts for type safety
3. Display tool execution - Show loading states and results to users
4. Manage history - Consider message persistence for long conversations
5. Implement retry logic - Provide easy way to retry failed requests

Generative UI

1. Separate concerns - Keep tools, components, and chat logic separate
2. Type tool outputs - Ensure component props match tool output schema
3. Handle loading states - Show input-streaming and input-available states
4. Error boundaries - Handle tool execution errors gracefully
5. Progressive enhancement - Start simple, add complexity incrementally

General

1. Use pnpm - Default package manager for all AI SDK projects
2. Pin versions - AI SDK 6 is beta; consider pinning versions for stability
3. Test edge cases - Test with various inputs, tool combinations, error scenarios
4. Monitor performance - Track token usage, execution time, and error rates
5. Iterate on prompts - System instructions have significant impact on results

---

Framework Support

AI SDK UI supports:

• React (@ai-sdk/react) - Full support: useChat, useCompletion, useObject
• Vue.js (@ai-sdk/vue) - useChat, useCompletion, useObject
• Svelte (@ai-sdk/svelte) - useChat, useCompletion, useObject
• Angular (@ai-sdk/angular) - useChat, useCompletion, useObject
• SolidJS (community) - Contributions welcome

---

Resources

• AI SDK 6 Docs
• Cookbook Examples
• Showcase
• GitHub
• Discussions
• Playground