KITE (Knowledge Intelligence for Technical Environments)

KITE is an AI-powered industrial knowledge platform that ingests maintenance logs, SOPs, work orders, and P&IDs into a Neo4j knowledge graph and Qdrant vector store, then exposes GraphRAG search, RCA, lessons-learned clustering, and compliance auditing through a React dashboard.

---

Architecture

┌─────────────┐     ┌──────────────┐     ┌─────────┐
│   React UI  │────▶│  FastAPI API │────▶│  Neo4j  │
│  (Vite/Nginx)│     │  + Agents    │     └─────────┘
└─────────────┘     │              │     ┌─────────┐
                    │              │────▶│ Qdrant  │
                    │              │     └─────────┘
                    │              │     ┌───────────┐
                    └──────────────┘────▶│ PostgreSQL│
                                         └───────────┘

Agents: Copilot (GraphRAG query), RCA, Lessons-Learned clustering, Compliance audit.

---

Prerequisites

• Docker & Docker Compose
• Python 3.12+
• Node.js 20+
• uv (Python package manager)

---

Quick Start (Docker)

1. Copy environment template and set your API key:

cp .env.example .env
# Edit .env — set GEMINI_API_KEY (required for LLM features)

2. Start the full stack:

docker compose up -d --build

3. Open the app:
   • Frontend: http://localhost:5173
   • API docs: http://localhost:8000/docs
   • Neo4j browser: http://localhost:7474

---

Local Development

Backend

cd backend
cp ../.env.example .env   # or use root .env
uv venv && uv pip install -r pyproject.toml
uv run uvicorn app.main:app --reload

API runs at http://localhost:8000.

Frontend

cd frontend
npm install
npm run dev

UI runs at http://localhost:5173. Vite proxies /api, /health, /metrics, and /version to the backend.

---

Environment Variables

See .env.example for the full list. Key variables:

Variable	Description	Default
GEMINI_API_KEY	Google Gemini API key (required for LLM)	—
LLM_PROVIDER	LLM backend: gemini, openai, groq, ollama	gemini
NEO4J_URI	Neo4j bolt URI	bolt://localhost:7687
DATABASE_URL	PostgreSQL async connection string	local docker default
QDRANT_URL	Qdrant HTTP endpoint	http://localhost:6333
CORS_ORIGINS	Comma-separated allowed origins	http://localhost:5173
VITE_WORKSPACE_NAME	Workspace label in UI	My Workspace

---

API Endpoints

Method	Path	Description
GET	/health	Service connectivity status
GET	/metrics	System counters
POST	/api/v1/ingest	Upload a document
GET	/api/v1/uploads	List past uploads
GET	/api/v1/review-queue	Entity resolution review items
POST	/api/v1/query	GraphRAG search / Copilot
GET	/api/v1/graph/nodes	Graph explorer data
GET	/api/v1/agents/rca/{id}	Root cause analysis
POST	/api/v1/agents/lessons/cluster	Lessons-learned clustering
GET	/api/v1/agents/compliance	Compliance audit

Legacy /api/... routes (without /v1) are also supported for backward compatibility.

---

Demo Flow

1. Ingest — Upload sample docs from Data_feed/synthetic_maintenance_docs/.
2. Overview — Confirm graph nodes and vectors are increasing.
3. Search / Copilot — Ask questions like "What caused the pump failure on P-101?"
4. Graph Explorer — Visualize equipment, failures, and relationships.
5. RCA — Enter an equipment tag (e.g. P-101) for an AI-written report.
6. Lessons-Learned — Run clustering to link similar failure modes.
7. Compliance — Review audit results and confirm/dismiss gaps.
8. Review Queue — Confirm entity alias matches flagged during ingestion.

---

Testing

cd backend
uv pip install -r pyproject.toml
pytest tests/ -v --timeout=60

---

Deployment Notes

• The frontend Docker image uses nginx and proxies /api, /health, /metrics, and /version to the backend container.
• Set CORS_ORIGINS to include your production frontend URL.
• All database volumes (neo4j_data, pg_data, qdrant_data) persist across restarts.
• Services use restart: unless-stopped in docker-compose.

---

Automated Benchmark Evaluation

cd backend
uv run python scripts/run_evaluation.py