Architecture
Understand how the AI Workflow Framework is structured and how components interact.
Architecture Overview
The framework follows a layered architecture with clear separation of concerns:
Text
┌─────────────────────────────────────────────────────────────┐
│ PRESENTATION LAYER │
│ Portal Wizards │ Backend Views │ API Endpoints │ Docs │
├─────────────────────────────────────────────────────────────┤
│ WORKFLOW LAYER │
│ Website │ Ebook │ Course │ Survey │ Video │ Custom... │
├─────────────────────────────────────────────────────────────┤
│ CORE ENGINE │
│ Templates │ Stages │ Tasks │ Queue │ Approvals │ Contracts │
├─────────────────────────────────────────────────────────────┤
│ AI PROVIDER LAYER │
│ OpenAI │ Anthropic │ Custom Providers │ Rate Limiting │
├─────────────────────────────────────────────────────────────┤
│ ODOO PLATFORM │
│ ORM │ Security │ Website │ Project │ Queue Job │
└─────────────────────────────────────────────────────────────┘Module Layers
Layer 1: Core Framework
| Module | Purpose |
|---|---|
ai_project_workflow |
Core workflow engine, templates, stages, task processing |
ai_project_workflow_portal |
Portal interface, wizards, documentation |
ai_project_workflow_api |
REST API endpoints for external integrations |
Layer 2: Workflow Implementations
| Module | Purpose |
|---|---|
ai_workflow_website |
Website creation workflow |
ai_workflow_ebook_creator |
Ebook creation workflow |
ai_workflow_course |
Course creation workflow |
ai_workflow_survey |
Survey creation workflow |
ai_workflow_video_creator |
Video creation workflow |
Core Models
Text
ai.project.template # Defines workflow structure
└── ai.project.stage # Individual stages in workflow
└── ai.template # AI prompt templates
ai.provider # AI provider configuration (OpenAI, etc.)
project.project # Extended with AI workflow fields
└── project.task # Extended with AI processing statesSee the API Reference for complete model documentation.
Execution Flow
Text
1. User creates project (Portal/Backend)
↓
2. System loads template & stages
↓
3. First stage executes (Discovery)
↓
4. AI Provider called via queue job
↓
5. Result stored, stage advances
↓
6. If approval required → Wait for user
↓
7. User approves → Next stage
↓
8. Repeat until complete