floriandomgjoni/laravel-ai-factory

Laravel package for generating realistic factory data using AI (OpenAI, Gemini, DeepSeek) with support for manual and hardcoded fields, model events, and bulk inserts.

dev-main 2025-06-03 18:02 UTC

Requires

• php: ^8.2
• illuminate/contracts: ^10.0||^11.0||^12.0
• spatie/laravel-package-tools: ^1.16

Requires (Dev)

• larastan/larastan: ^2.9||^3.0
• laravel/pint: ^1.14
• nunomaduro/collision: ^8.1.1||^7.10.0
• orchestra/testbench: ^10.0.0||^9.0.0||^8.22.0
• pestphp/pest: ^3.0
• pestphp/pest-plugin-arch: ^3.0
• pestphp/pest-plugin-laravel: ^3.0
• phpstan/extension-installer: ^1.3||^2.0
• phpstan/phpstan-deprecation-rules: ^1.1||^2.0
• phpstan/phpstan-phpunit: ^1.3||^2.0

MIT

• Florian Domgjoni

• laravel
• laravel-ai-factory

README

Laravel AI Factory is a developer-friendly package for generating realistic test data using AI models. It integrates seamlessly with Laravel factories and supports both AI-generated and manually defined fields, with the option to use bulk or individual inserts.

🚀 Features

• ✅ Generate factory data using AI prompts
• ✅ Supports OpenAI - Gemini and DeepSeek support coming soon...
• ✅ Mix AI-generated fields with hardcoded or Faker-generated fields
• ✅ Optional bulk insert for performance (without triggering model events)
• ✅ Easy to extend and override

📦 Installation

composer require floriandomgjoni/laravel-ai-factory

⚙️ Configuration

Publish the config:

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

Update your .env with the appropriate API keys:

AI_FACTORY_DRIVER=openai
AI_FACTORY_OPENAI_API_KEY=your-api-key
AI_FACTORY_OPENAI_MODEL=gpt-4o-mini

or, for a local LLM:

AI_FACTORY_DRIVER=local
AI_FACTORY_LOCAL_URL=http://localhost:8080/v1/chat/completions
AI_FACTORY_LOCAL_API_KEY=""
AI_FACTORY_LOCAL_MODEL=Default

🧠 Defining AI Fields

In your factory, use the HasAIFactory trait and define the aiFields() method:

<?php
use FlorianDomgjoni\AIFactory\Concerns\HasAIFactory;

class UserFactory extends Factory
{
    use HasAIFactory;

    protected $model = \App\Models\User::class;

    public function aiFields(): array
    {
        return [
            'name' => 'Generate a realistic full name',
            'email' => fn () => fake()->unique()->safeEmail(),
            'password' => fn () => bcrypt('password'),
            'api_token' => fn () => Str::random(60),
        ];
    }
}

• Use a string for AI-generated fields (prompt).
• Use a callable for manual or faker-based fields.

You also need to fill the $model property with the model class inside the factory. It is required in order to function properly.

🛠 Usage

Basic Usage

User::factory()->count(5)->createWithAI();

Override Fields

User::factory()->count(3)->createWithAI([
    'email' => fn () => fake()->unique()->safeEmail(),
    'role' => fn () => 'admin'
]);

Bulk Insert (faster, skips model events)

User::factory()->count(100)->createWithAI([], true);

🧪 Example Prompt

[
    'title' => 'Generate a blog post title related to technology',
    'content' => 'Generate a paragraph of blog content about AI',
    'published_at' => fn () => now()->subDays(rand(1, 30)),
]

📂 Config File

return [
    'driver' => env('AI_FACTORY_DRIVER', 'openai'),

    'openai' => [
        'api_key' => env('AI_FACTORY_OPENAI_API_KEY'),
        'model' => env('AI_FACTORY_OPENAI_MODEL', 'gpt-4o-mini'),
    ],

    'local' => [
        'url' => env('AI_FACTORY_LOCAL_URL', 'http://localhost:8080'),
        'api_key' => env('AI_FACTORY_LOCAL_API_KEY', null),
        'model' => env('AI_FACTORY_LOCAL_MODEL', null),
    ],
];

🛡 Error Handling

• AI data generation is wrapped in try/catch.
• Model creation errors are logged individually.
• Invalid JSON or failed API calls will throw descriptive exceptions.

📄 License

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