AI Service Package

This package provides a unified interface to interact with various AI services, including OpenAI, Gemini and Claude. It is designed for Laravel applications and offers an extensible and configurable solution for AI-powered functionality such as text generation, embeddings, and more.

---

Features

• Multi-Provider Support: Interact with OpenAI, Gemini, and Claude.
• Centralized Configuration: Manage settings from a single config/ai.php file.
• Extensible Interface: Add support for new AI providers with ease.
• Error Handling: Includes try...catch blocks to handle API errors gracefully.
• Laravel Compatible: Designed to work seamlessly with Laravel’s dependency injection and service container.

---

Installation

Step 1: Require the Package

composer require kwakuofosuagyeman/ai-assistant

Step 2: Publish Configuration File

Publish the configuration file to customize settings:

php artisan vendor:publish --tag=ai-config

This will create a config/ai.php file in your Laravel application.

Step 3: Set Environment Variables

Add your API keys and configuration to the .env file:

YOUR_PROVIDER_API_KEY=your-provider-api-key

---

Configuration

The config/ai.php file contains all the settings for the supported AI providers. Example:

return [
    'default' => 'openai',
    'providers' => [
        'services' => [
            'openai' => [
                'api_key' => env('OPENAI_API_KEY'),
                'default_model' => 'text-davinci-003',
                'chat_model' => 'gpt-4',
                'default_max_tokens' => 150,
                'embedding_model' => 'text-embedding-ada-002',
                'default_temperature' => 0.7,
            ],
            'claude' => [
                'api_key' => env('CLAUDE_API_KEY'),
                'base_url' => 'https://api.anthropic.com/v1/messages/',
                'default_max_tokens' => 1024,
                'batch_url' => 'https://api.anthropic.com/v1/messages/batches/',
                'model' => 'claude-3-5-sonnet-20241022',
                'version' => env('CLAUDE_API_VERSION'),

            ],
            'gemini' => [
                'api_key' => env('GEMINI_API_KEY'),
                'base_url' => 'https://generativelanguage.googleapis.com/v1beta/models/',
                'default_model' => 'gemini-2.0-flash:generateContent',
            ]
        ],
    ]
    
];

---

Usage

Method 1: Use the Service in Your Application

Use the required service where needed, such as in controllers or other services:

use Kwakuofosuagyeman\AIAssistant\Services\OpenAIService;

class ExampleController extends Controller
{
    protected $aiManager;


    public function generateText(Request $request)
    {
        $prompt = $request->input('prompt');

        try {
            $openAIService = new OpenAIService();
            $response = $openAIService->generateText("Write a poem about the sea.");

            return response()->json($response);
        } catch (\Exception $e) {
            return response()->json(['error' => $e->getMessage()], 500);
        }
    }
}

Method 2: Use the Service in Your Application (If you wish to use multiple AI providers)

Inject the AIService interface where needed, such as in controllers or other services:

use Kwakuofosuagyeman\AIAssistant\AIManager;

class ExampleController extends Controller
{
    protected $aiManager;

    public function __construct(AIManager $aiManager)
    {
        $this->aiManager = $aiManager;
    }

    public function generateText(Request $request)
    {
        $provider = $request->input('provider'); // Optional, defaults to config('ai.default')
        $prompt = $request->input('prompt');

        try {
            $response = $this->aiManager->resolveService($provider)->generateText($prompt);

            return response()->json($response);
        } catch (\Exception $e) {
            return response()->json(['error' => $e->getMessage()], 500);
        }
    }
}

---

Supported Methods

OpenAI

This system uses openai-php/client to aid in making requests to openai

generateText(string $prompt, array $options = []): array

Generates text based on the given prompt.

$response = $aiService->resolveService('openai')->generateText('Tell me a story about space exploration.', [
    'temperature' => 0.8,
    'max_tokens' => 200,
]);

echo $response['data'];

generateEmbeddings(string $text): array

Generates embeddings for the given text. This uses the text-embedding-ada-002 by default.

$response = $aiService->resolveService('openai')->generateEmbeddings('Artificial Intelligence');

print_r($response['data']);

chat(array $messages, array $options = []): array

Facilitates a conversation-like interaction with the AI model.

$messages = [
    ['role' => 'user', 'content' => 'What is the capital of France?'],
    ['role' => 'assistant', 'content' => 'The capital of France is Paris.'],
];

$response = $aiService->resolveService('openai')->chat($messages);

echo $response['data']['content'];

analyzeSentiment(string $text, array $options = []): array

Analyze the sentiment of a text. The labels given to the ai are ['negative', 'neutral', 'positive']

$text = 'Text to analyze';

$response = $aiService->resolveService('openai')->analyzeSentiment($text);

summarizeText(string $text, array $options = []): array

Transcribes audio files.

$text = 'Text to summarize';

$response = $aiService->resolveService('openai')->summarizeText($text);

translateText(string $text, string $targetLanguage, array $options = []): array

Translates text from its current language to the targetLanguage.

$text = 'Text to translate';
$targetLangugae = 'French';

$response = $aiService->resolveService('openai')->translateText($text, $targetLanguage);

generateCode(string $prompt, array $options = []): array

Generates code. This has a default temperature of 0.2

$text = 'Generate python code to reverse a string';


$response = $aiService->resolveService('openai')->generateCode($prompt);

Gemini

generateText(string $prompt, array $options = []): array

Generates text based on the given prompt.

$response = $aiService->resolveService('gemini')->generateText('Tell me a story about space exploration.');

echo $response['data'];

transcribeAudio(string $audioPath, array $options = []): array

Transcribes the given audio into text.

$audio = "path/to/audio";


$response = $aiService->resolveService('gemini')->transcribeAudio($audio);

Claude

Generate Text

Use the generateText method to generate AI responses based on a text prompt.

use Kwakuofosuagyeman\AIAssistant\Services\ClaudeAIService;

$aiService = new ClaudeAIService();

$prompt = "Write a short story about a brave cat.";
$options = ['maxTokens' => 150];

$response = $aiService->generateText($prompt, $options);

if (isset($response['error'])) {
    echo "Error: " . $response['error'];
} else {
    echo "Generated Text: " . $response['content'];
}

Analyze a Document with a Query

Analyze a document (e.g., a PDF file) and provide answers to a query about its content.

$fileUrl = 'https://example.com/path-to-document.pdf';
$options = ['query' => 'Summarize the main points of this document.'];

$response = $aiService->analyzeDocumentWithQuery($fileUrl, $options);

if (isset($response['error'])) {
    echo "Error: " . $response['error'];
} else {
    echo "Analysis Result: " . json_encode($response);
}

Use Tool with Messages

Send specific messages to the Claude AI service along with tools for interaction. This can be used for chatbots,

$messages = [
    ['role' => 'user', 'content' => 'What is the weather like in San Francisco']
];
$options = [
    'tool' => [
      {
        "name": "get_weather",
        "description": "Get the current weather in a given location",
        "input_schema": {
          "type": "object",
          "properties": {
            "location": {
              "type": "string",
              "description": "The city and state, e.g. San Francisco, CA"
            }
          },
          "required": ["location"]
        }
      }
    ],
    'maxTokens' => 1024
];

$response = $aiService->useTool($messages, $options);

if (isset($response['error'])) {
    echo "Error: " . $response['error'];
} else {
    echo "Tool Response: " . json_encode($response);
}

Generate Batch Messages

Send multiple prompts as a batch and receive responses.

$batches = [
        {
            "custom_id": "my-first-request",
            "params": {
                "model": "claude-3-5-sonnet-20241022",
                "max_tokens": 1024,
                "messages": [
                    {"role": "user", "content": "Hello, world"}
                ]
            }
        },
        {
            "custom_id": "my-second-request",
            "params": {
                "model": "claude-3-5-sonnet-20241022",
                "max_tokens": 1024,
                "messages": [
                    {"role": "user", "content": "Hi again, friend"}
                ]
            }
        }
    ];

$response = $aiService->generateBatchMessages($batches);

if (isset($response['error'])) {
    echo "Error: " . $response['error'];
} else {
    echo "Batch Responses: " . json_encode($response);
}

Manage Batch Requests

List Batch Requests:

$response = $aiService->listBatchMessages();
echo json_encode($response);

Retrieve a Batch Result:

$token = 'batch-id';
$response = $aiService->getBatchMessagesResult($token);
echo json_encode($response);

Cancel a Batch Request:

$token = 'batch-id';
$response = $aiService->cancelMessageBatch($token);
echo json_encode($response);

---

Adding a New AI Provider

To add support for a new AI provider:

1. Create a New Service: Implement the AIService interface:

   namespace Kwakuofosuagyeman\AIAssistant\Services;
   
   use Kwakuofosuagyeman\AIAssistant\Contracts\AIService;
   
   class NewAIProviderService implements AIService
   {
       public function generateText(string $prompt, array $options = []): array
       {
           // Implement text generation
       }
   
       // Implement other methods as needed
   }

2. Register the New Service: Bind the new service in your service provider.

---

Testing

You can test the service using PHPUnit. Example test for the OpenAI integration:

use Tests\TestCase;
use Kwakuofosuagyeman\AIAssistant\Contracts\AIService;

class OpenAIServiceTest extends TestCase
{
    public function testGenerateText()
    {
        $aiService = $this->app->make(AIService::class);
        $response = $aiService->resolveService('openai')->generateText('Test prompt');

        $this->assertArrayHasKey('data', $response);
    }
}

---

Contributing

Contributions are welcome! Please follow the standard Laravel and PHP development guidelines.

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Submit a pull request with a clear description of your changes.

---

License

This package is open-sourced software licensed under the MIT license.

---

Support

For issues, please open an issue on the GitHub repository or contact the maintainers.