An enterprise AI assistant capable of answering natural language inquiries regarding building assets, infrastructure, operational status, telemetry health, and technical documentation by unifying three distinct knowledge tiers.
The Asset Intelligence Platform combines three decoupled data sources into a single retrieval & reasoning pipeline:
User Natural Language Query
│
▼
LLM Orchestrator
┌───────────────────────┼───────────────────────┐
▼ ▼ ▼
Neo4j Graph Telemetry Engine PDF Document Store
(Relationships & (Historical Time-Series (Technical Manuals &
Metadata Layer) Metrics) RAG Vector DB)
- Knowledge Graph (Neo4j): Stores semantic topology, facility hierarchies, equipment components, sensor metadata, warranties, contracts, and technicians.
- Telemetry Engine (CSV / Pandas): Manages high-frequency historical sensor readings (e.g., Supply/Return Air Temperature, Fan RPM, Energy Consumption).
- Technical Document Store (RAG): Stores unstructured PDF manuals, service records, and engineering schematics for semantic vector search.
- LLM Orchestrator: Fuses context across all three data tiers into a factual, context-aware answer.
AssetIntelligence/
│
├── README.md # Project documentation & setup guide
├── requirements.txt # Python package dependencies
├── .env.example # Environment variables template
├── main.py # Main CLI entry point
├── config.py # Centralized configuration management
│
├── graph/ # Neo4j Knowledge Graph interface module
│ ├── __init__.py
│ ├── neo4j_client.py # Neo4j connection pool and query execution client
│ └── graph_queries.py # Cypher query templates and retrieval methods
│
├── telemetry/ # CSV Telemetry processing engine
│ ├── __init__.py
│ ├── telemetry_reader.py # Historical telemetry loading and filtering
│ └── telemetry_analysis.py # Statistical aggregation and anomaly detection
│
├── documents/ # Technical document loading & RAG module
│ ├── __init__.py
│ └── document_loader.py # PDF document parsing and text chunking
│
├── llm/ # Multi-source LLM Orchestration
│ ├── __init__.py
│ └── orchestrator.py # Context fusion, prompt construction, and response synthesis
│
└── utils/ # Shared utility modules
├── logger.py # Centralized logging configuration
└── exceptions.py # Custom application exception hierarchy
-
Clone & Set Up Environment:
python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install -r requirements.txt
-
Configure Environment Variables: Copy
.env.exampleto.envand populate your database credentials and API keys:cp .env.example .env
-
Run Application:
python main.py