Multi-agent AI system for Governance, Risk & Compliance (GRC)
Grounded in your own documents — not generic AI knowledge.
RegulAIte is an open-source GRC assistant built by students from OteriaCyberSchool. It lets organizations upload their internal documents (policies, standards, contracts) and query them through a conversational AI — powered by a multi-agent architecture and a RAG (Retrieval-Augmented Generation) pipeline.
Important
Instead of relying on generic AI knowledge, RegulAIte answers compliance questions based only on your documents, making responses traceable and auditable.
| Feature | Description | |
|---|---|---|
| 🤖 | Multi-agent orchestration | An Orchestrator delegates tasks to specialized agents (compliance analysis, gap analysis, risk assessment, governance) |
| 🔍 | RAG pipeline | Documents are chunked, embedded, and stored in a vector database (Qdrant) for precise retrieval |
| 📄 | Document management | Upload and parse PDF, Word, and other formats via the Unstructured API |
| ⚡ | Async task queue | Long-running agent tasks handled by Celery + Redis, keeping the UI responsive |
| 🔐 | Authentication | User and organization management built-in |
| 🌍 | Bilingual support | French and English query handling |
flowchart TD
A7["Chat (Frontend)"]
A9["Documents (Frontend)"]
A8["Auth"]
A6["API Routing (FastAPI)"]
A5["Task Queue (Celery/Redis)"]
A1["Orchestrator Agent"]
A0["Specialized Agents"]
A4["Tool Registry"]
A2["RAG System (Qdrant)"]
A3["Document Chunks"]
A7 -- "Sends requests" --> A6
A9 -- "Uploads documents" --> A6
A8 -- "Protects" --> A6
A6 -- "Routes to" --> A1
A6 -- "Queues tasks" --> A5
A5 -- "Runs" --> A0
A5 -- "Processes" --> A3
A1 -- "Delegates to" --> A0
A0 -- "Uses" --> A4
A0 -- "Queries" --> A2
A4 -- "Queries" --> A2
A2 -- "Retrieves" --> A3
| Layer | Technology |
|---|---|
| 🖥️ Frontend | React |
| ⚙️ Backend | Python / FastAPI |
| 🗄️ Vector DB | Qdrant |
| 🗃️ Relational DB | MariaDB |
| 📬 Task queue | Celery + Redis |
| 📑 Document parsing | Unstructured |
| 🐳 Containerization | Docker Compose |
- Docker and Docker Compose
- An LLM API key (OpenAI, Anthropic, or compatible)
git clone git@github.com:HXLLO/RegulAIte.git
cd RegulAIte
# Copy and configure environment variables
cp backend/.env.example backend/.env
# Edit backend/.env with your API keys and credentials
docker compose up --buildOnce running, access the services:
| Service | URL |
|---|---|
| 🌐 Frontend | http://localhost:3000 |
| 🔌 Backend API | http://localhost:8000 |
| 📖 API docs | http://localhost:8000/docs |
| 📊 Celery monitor | http://localhost:5555 |
| 🔷 Qdrant dashboard | http://localhost:6333/dashboard |
RegulAIte/
├── backend/
│ ├── agent_framework/ # Agents, orchestrator, tools, RAG
│ ├── api/ # FastAPI routes
│ ├── queuing_sys/ # Celery workers
│ └── config/ # Service configurations
├── front-end/ # React application
├── database-backups/ # Qdrant snapshots
├── docs/ # Technical documentation
└── docker-compose.yml
Detailed technical documentation is available in /docs:
Built by students at OteriaCyberSchool — French cybersecurity engineering school.