API FlutterAI

API Reference: FlutterAI

The main class for interacting with AI providers.

Constructor

FlutterAI({
  required AIProvider provider,
  required AIConfig config,
  ContextManager? contextManager,
})

Parameters

Parameter	Type	Required	Description
provider	AIProvider	✅	The AI provider to use
config	AIConfig	✅	Configuration options
contextManager	ContextManager?	❌	Custom context manager

Example

final ai = FlutterAI(
  provider: AIProvider.openai,
  config: AIConfig(
    apiKey: 'your-api-key',
    model: 'gpt-5.5',
    temperature: 0.7,
    systemPrompt: 'You are a helpful assistant.',
  ),
);

---

Properties

provider

AIProvider get provider

The type of AI provider being used.

config

AIConfig get config

The current configuration.

context

ContextManager get context

The context manager instance.

conversation

Conversation get conversation

The current conversation history.

---

Methods

chat

Sends a simple text message.

Future<AIResponse> chat(
  String message, {
  bool addToContext = true,
})

Parameter	Type	Default	Description
message	String	-	The message to send
addToContext	bool	true	Whether to add to conversation history

Returns: Future<AIResponse>

final response = await ai.chat('What is Flutter?');
print(response.text);
print(response.usage?.totalTokens);

---

chatWithContent

Sends multimodal content (text, images, etc.).

Future<AIResponse> chatWithContent(
  List<Content> content, {
  bool addToContext = true,
})

Parameter	Type	Default	Description
content	List<Content>	-	The content to send
addToContext	bool	true	Whether to add to history

final response = await ai.chatWithContent([
  const TextContent('What is in this image?'),
  const ImageContent.fromUrl('https://example.com/image.jpg'),
]);

---

chatWithTools

Sends a message with function/tool support.

Future<AIResponse> chatWithTools(
  String message, {
  required List<Tool> tools,
  ToolChoice? toolChoice,
  bool addToContext = true,
})

Parameter	Type	Default	Description
message	String	-	The message to send
tools	List<Tool>	-	Available tools
toolChoice	ToolChoice?	null	How to use tools
addToContext	bool	true	Add to history

final response = await ai.chatWithTools(
  'What is the weather in Paris?',
  tools: [weatherTool],
);

if (response.hasToolCalls) {
  // Handle tool calls
}

---

submitToolResult

Submits the result of a tool call back to the AI.

Future<AIResponse> submitToolResult({
  required String toolCallId,
  required String name,
  required dynamic result,
  bool isError = false,
})

Parameter	Type	Default	Description
toolCallId	String	-	ID from the tool call
name	String	-	Tool name
result	dynamic	-	Tool execution result
isError	bool	false	Whether result is an error

final finalResponse = await ai.submitToolResult(
  toolCallId: 'call_123',
  name: 'get_weather',
  result: {'temperature': 22, 'condition': 'sunny'},
);
print(finalResponse.text);

---

streamChat

Streams a response for a text message.

Stream<StreamChunk> streamChat(
  String message, {
  bool addToContext = true,
})

await for (final chunk in ai.streamChat('Tell me a story')) {
  if (chunk.isDelta) {
    stdout.write(chunk.delta);
  }
  if (chunk.isDone) {
    print('\nTokens: ${chunk.usage?.totalTokens}');
  }
}

---

streamChatWithContent

Streams a response for multimodal content.

Stream<StreamChunk> streamChatWithContent(
  List<Content> content, {
  bool addToContext = true,
})

---

streamChatWithTools

Streams a response with tool support.

Stream<StreamChunk> streamChatWithTools(
  String message, {
  required List<Tool> tools,
  ToolChoice? toolChoice,
  bool addToContext = true,
})

---

clearContext

Clears the conversation history.

void clearContext()

ai.clearContext();
// Conversation is now empty (system prompt retained)

---

dispose

Releases resources.

void dispose()

// Always call when done
ai.dispose();

---

Complete Example

import 'package:flutter_ai_sdk/flutter_ai_sdk.dart';

Future<void> main() async {
  // Initialize
  final ai = FlutterAI(
    provider: AIProvider.openai,
    config: AIConfig(
      apiKey: 'your-api-key',
      model: 'gpt-5.5',
      temperature: 0.7,
      maxTokens: 1000,
      systemPrompt: 'You are a helpful assistant.',
    ),
  );

  try {
    // Simple chat
    final response1 = await ai.chat('Hello!');
    print('AI: ${response1.text}');

    // Multimodal chat
    final response2 = await ai.chatWithContent([
      const TextContent('Describe this:'),
      const ImageContent.fromUrl('https://example.com/image.jpg'),
    ]);
    print('AI: ${response2.text}');

    // Streaming
    print('\nStreaming response:');
    await for (final chunk in ai.streamChat('Tell me a joke')) {
      if (chunk.isDelta) stdout.write(chunk.delta);
    }
    print();

    // With tools
    final response3 = await ai.chatWithTools(
      'What is 25 * 4?',
      tools: [calculatorTool],
    );
    
    if (response3.hasToolCalls) {
      for (final call in response3.toolCalls!) {
        final result = await executeCalculation(call.arguments);
        final finalResponse = await ai.submitToolResult(
          toolCallId: call.id,
          name: call.name,
          result: result,
        );
        print('AI: ${finalResponse.text}');
      }
    }

    // Check context
    print('\nContext: ${ai.conversation.length} messages');
    print('Tokens: ${ai.context.estimatedTokens}');

  } finally {
    ai.dispose();
  }
}

Related

• AIConfig - Configuration options
• AIResponse - Response structure
• Configuration - Configuration guide