About Shopfloor Copilot
AI-Powered Manufacturing Intelligence — On Your Terms
Live Demo & Pilot Evaluation Phase. Shopfloor Copilot is an on-premise AI-powered MES by Machina Intelligence. The public hosted instance runs on simulated plant data and uses a cloud LLM for speed. On-premise pilot deployments run locally via Docker Compose and Ollama. Production readiness must be validated during pilot evaluation.
What We Build
Shopfloor Copilot is a full-stack Manufacturing Execution System (MES) designed for discrete manufacturers who want real-time OEE monitoring, AI-assisted diagnostics, and deep machine connectivity — without surrendering their production data to the cloud.
We combine a modern Python/FastAPI backend with NiceGUI dashboards, PostgreSQL time-series storage, and local LLM inference via Ollama, so your AI stays inside your network. No subscription to an external AI API. No production data leaving the factory floor.
Core Capabilities
- OEE Monitoring Dashboard — live availability, performance, quality metrics per line and shift
- OPC UA Explorer & Semantic Signals — connect to any PLC or SCADA system via OPC UA, map signals visually
- Predictive Maintenance — ML-based equipment health scoring, alert thresholds, maintenance scheduling
- AI Diagnostics (RAG) — operators ask natural language questions; AI answers from plant knowledge base
- Digital Shift Handover — structured handover notes, issue tracking, team collaboration
- Andon Board — live alert board for supervisors and line managers
- Visual Line Designer — BPMN-based shopfloor layout editor with station management
- KPI Engine — 30+ configurable manufacturing KPIs with formula parser
Our Approach
Most MES vendors push cloud-first architectures. We believe manufacturers — especially those in regulated industries (automotive, pharma, aerospace) — need a different option: one where all data stays on-premise, where the AI runs locally, and where the software can operate in air-gapped environments.
Shopfloor Copilot runs entirely in Docker containers on a single server. No external services, no vendor lock-in, no metered API calls. You own the stack.
MachinaOS Integration
The MCP interface exposes manufacturing context and tools to MachinaOS for governed multi-agent orchestration. Pilot evaluation should validate permissions, approvals, and audit trails for the intended workflows.
Technology Stack
- Frontend: NiceGUI 2.5 (Python → Quasar/Vue.js)
- Backend: FastAPI 24+ API routers, SQLAlchemy 2.0
- Database: PostgreSQL 16
- AI/ML: Ollama (local LLM), ChromaDB (vector store for RAG)
- Connectivity: OPC UA via asyncua, MQTT-ready
- Deployment: Docker Compose, nginx reverse proxy
Contact
For product enquiries, partnership requests, or technical questions:
Or use our demo and pilot enquiry instructions.