Backend Overview

The TKM AI Agency Platform backend is a sophisticated FastAPI-based system that orchestrates multiple AI agents through an event-driven architecture. The system is organized into several key components:

Directory Structure

Backend/
├── AI_Agency/          # Main application directory
│   ├── Backoffice/    # Internal operations and API endpoints
│   ├── CRM/           # Customer interaction management
│   ├── Core_Components/ # System infrastructure
│   └── Configuration/ # System configuration
├── main.py            # Application entry point
├── requirements.txt   # Project dependencies
└── logs/             # System logs

Core Architecture

FastAPI Application

The system is built on FastAPI, providing:

• High-performance async/await support
• Automatic OpenAPI documentation
• Built-in CORS middleware
• Static file serving for media content

Agent System

Backoffice Agents

Handle internal operations and API endpoints:

• Input Processing
  • Burchelli: Text processing and analysis
  • Rufa: Audio processing and transcription
  • Scalaris: Document management and processing
  • Hova: Chat response management
• System Management
  • Atta: Context and state management
  • Bala: Embedding generation and processing
  • Barbatus: Session management and authentication
  • Bauri: System utilities and support

CRM Agents

Manage customer interactions and business logic:

• Core Processing

  • Prometheus: Main response generation
  • Invicta: Request routing and orchestration
  • Humile: Translation and language services
  • Lyra: Permission and access management

• Specialized Processing

  • Carina: Specialized interactions
  • Orfeo: Custom processing
  • Orion: System monitoring
  • Pegaso: Advanced features

Core Infrastructure

Event System

• Asynchronous event bus for inter-agent communication
• Event-driven architecture for decoupled operations
• Centralized event handling and distribution

Mediator Service

• Agent registration and management
• Message routing between agents
• State coordination and synchronization

Configuration Management

• Centralized configuration system
• Environment-specific settings
• Agent-specific configurations
• LLM provider settings

API Structure

Public Endpoints

/api/
├── burchelli/  # Text processing
├── rufa/       # Audio processing
├── hova/       # Chat management
├── scalaris/   # Document management
├── barbatus/   # Session management
├── humile/     # Translation services
└── orfeo/      # Custom processing

System Features

• Health monitoring (/health)
• Static file serving for media content
• Comprehensive error handling and logging
• Request validation and sanitization

Data Management

• Structured logging with rotation
• File storage for documents and media
• In-memory caching for performance
• Persistent storage for critical data

Security Features

• CORS protection
• Session management
• Permission-based access control
• Environment variable management