Local LLM
Qwen3.6 35B running on dedicated GPU inference server via llama.cpp
pdfplumber
Extracts text from digital PDF manuals and catalogs
Tesseract OCR
Reads scanned and image-based PDFs automatically
ChromaDB
Vector database for semantic search across all ingested documents
SearXNG
Private web search for live pricing and availability
sentence-transformers
GPU-accelerated document embeddings (all-MiniLM-L6-v2)
Cline / Pi Agent
AI coding agents for admin tasks via the Ork panel
PIN-Protected Admin
Secure orchestration panel for document and system management
FastAPI
Backend API server (24/7 uptime)
NAS Storage
Documents stored on network-attached storage (OMV)