Alea's Intelligence

Project Description

Alea's Intelligence is a dedicated application crafted to streamline communication with Artificial Intelligence. It's specifically designed to leverage APIs from various platforms, enabling seamless interaction with multiple AI models simultaneously. Currently, it integrates with OpenRouter AI and Groq, providing a versatile interface for diverse AI capabilities.

Notably, this application is built entirely from the ground up using native code, without the assistance of any external frameworks. This approach allows for a lean and focused implementation.

Key Features

• Multi-Model AI Communication
  Facilitates user interaction with different AI models concurrently, sending messages and receiving comprehensive responses.

• Audio Transcription
  Utilizes Whisper AI to convert spoken input into text, preparing it for further processing by the integrated AI models.

• Intelligent Memory System
  Manages conversation history and identifies pinned topics, ensuring that the AI receives relevant context for ongoing dialogues.

System Flow Diagrams

AI Communication Flow

This diagram illustrates how user messages are processed, dispatched to the AI, responses are received, and subsequently returned to the user interface. The flow incorporates robust input validation and mechanisms for handling request cancellations.

Audio Transcript Flow

This diagram outlines the comprehensive audio transcription process. It begins with audio input from the user, proceeds with transcription using Whisper AI, and then transmits the transcribed text to the AI model for further action.

Memory System Flow

This diagram details how the memory system efficiently manages chat history. It intelligently checks for existing conversational context, integrates the three most recent chat histories along with any pinned topics into the current message, and forwards this enriched data to the AI proxy.

---

System Requirements

---

Web Server Configuration (Nginx)

• client_max_body_size 50M: Configures the maximum permissible size for client request bodies, accommodating larger data transfers.
• Dedicated server blocks are implemented for enhanced security, utilizing distinct domains for public and private folders.
  • Public folder: Where users interact with the AI application
  • Private folder: Used for managing AI APIs and models

PHP Configuration

CLI and FPM (php.ini)

• upload_max_filesize = 50M: Sets the maximum allowed size for file uploads.
• post_max_size = 50M: Defines the maximum size for POST data, crucial for handling larger requests.

⚠️ Ensure that php-curl is properly installed for robust communication capabilities.

MySQL Configuration

• Database: intelligence_db
  Designated as the default database for the project, centralizing data storage.

• Installation of phpMyAdmin is highly recommended for efficient and user-friendly database management.

---

Whisper AI Installation

To set up Whisper AI for audio transcription, follow these steps:

1. Install System Dependencies:

sudo apt update && sudo apt install ffmpeg python3-pip python3-venv -y

2. Create a New Virtual Environment:

cd /var/www/
python3 -m venv whisper-env
source whisper-env/bin/activate

3. Upgrade Pip:

pip install --upgrade pip

4. Install Torch:

With CUDA (for GPU acceleration):

pip install torch torchvision torchaudio

Without CUDA (CPU-only):

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

5. Install Whisper AI from GitHub:

pip install git+https://github.com/openai/whisper.git

6. Grant Access Permissions for Whisper AI:

sudo chown -R www-data:www-data whisper-env
sudo chmod -R 755 whisper-env

7. Grant Permissions for the Application:

sudo chown -R www-data:www-data /var/www/aleas-intelligence
sudo chmod -R 755 aleas-intelligence

8. Test Installation

Upon completing the installation, install the base Whisper AI model and verify its functionality by transcribing an audio file into text before integrating it into the main application.

There are five models that can be installed: tiny, base, small, medium and large

whisper your-audio.mp3 --model base --language Indonesian --task transcribe --output_format txt

---

Usage

To set up and run Alea's Intelligence, follow these steps:

1. Place the Repository in Your Web Server Directory

Clone or copy this repository into your web server's root directory, commonly:

/var/www/aleas-intelligence/

Make sure the directory structure remains intact, especially the separation between public and private folders.

After placing the repository in your web server directory, configure the connect.php file with your database credentials to enable data connectivity.

2. Set Proper Permissions

Ensure that the server has the appropriate read/write permissions:

sudo chown -R www-data:www-data /var/www/aleas-intelligence
sudo chmod -R 755 /var/www/aleas-intelligence

3. Import the Database

Using phpMyAdmin or the MySQL CLI, import the provided SQL file (intelligence_db.sql) into your MySQL server:

mysql -u root -p intelligence_db < /path/to/intelligence_db.sql

Make sure the database intelligence_db is created before running the import.

4. Add API Credentials and AI Models via Admin Panel

Open the admin/private panel in your browser (e.g., http://yourdomain.com/private/) and log in (no authentication is required by default — you can implement it if needed).

From there:

• Navigate to the API Management section
• Add one or more APIs with the following required fields:
  • model_name (e.g., gpt-4, llama3-70b)
  • api_key (get this from OpenRouter or Groq)
  • url (base URL for the API provider)

Example for OpenRouter:

• model_name: mistralai/mixtral-8x7b
• api_key: sk-...
• url: https://openrouter.ai/api/v1/chat/completions

Example for Groq:

• model_name: llama3-70b-8192
• api_key: gsk_...
• url: https://api.groq.com/openai/v1/chat/completions

The system uses curl to communicate with the external APIs, so ensure all credentials are correct.

5. Configure PHP (If Not Yet Done)

Make sure your php.ini includes:

upload_max_filesize = 50M
post_max_size = 50M

Also make sure php-curl is installed:

sudo apt install php-curl

6. Configure Whisper AI

Once Whisper AI is installed, configure the voice_recognizer.php file by adjusting it to match the installation path and the model used.

7. Start Using the Application

Navigate to your public folder (e.g., http://yourdomain.com/) to interact with the AI system. You can:

• Send messages to different AI models
• Upload audio to be transcribed
• View context-aware replies thanks to the integrated memory system

Make sure your web server (e.g., Nginx) is properly serving both the public and private directories with appropriate security settings.

⚠️ For security, do not expose your private admin panel (/private/) to the public internet without restricting access.

---

You’re now ready to use Alea's Intelligence to interact with multiple AI models in a streamlined and dynamic environment!

Preview

Light Mode

Dark Mode

---

Contributing

This project is a labor of passion, shared with the community out of pure hobby interest. As such, there is currently no dedicated login portal implemented. If you are interested in contributing to this aspect or any other part of the project, your contributions are highly welcome!

Please note: I am sharing this project purely out of personal interest, and I am not a professional programmer—in fact, I don't even code. This project is a testament to what can be achieved through exploration and enthusiasm.

License

This project was created entirely out of personal passion. I am not a programmer, and truthfully, I might not even know how to code. You're completely free to use, modify, or distribute this project as you wish.

I genuinely hope you enjoy using it. If you do, feel free to reach out to me via Instagram DM. 😊