Insight Stack

•
4
•

Unifies an industrial AI stack for real-time machine insights, integrating LSTM anomaly detection, adaptive Bayesian optimization, and a local LLM assistant.

The Insight Stack is a comprehensive, lightweight industrial AI solution designed to provide real-time machine insights. It integrates an MCP gateway that exposes database-backed tools and a local LLM (llama.cpp) for natural language interaction, alongside an Insight backend performing LSTM-based anomaly detection and generating operator-facing explanations. Data is managed with a MySQL database for sensor schema, and an optional OPC UA simulator facilitates real-time data streaming. The system is built to offer robust integration for Human-Machine Interfaces, enabling proactive monitoring and intelligent operational support within industrial environments.

Características Principales

014 GitHub stars
02LSTM-AE based anomaly detection for real-time machine insights
03Local LLM assistant (llama.cpp) integrated via an MCP gateway for natural language interaction
04OPC UA simulator for generating streaming production machine data
05MySQL database for sensor data storage and LLM user provisioning
06Adaptive Bayesian optimization for process control and efficiency

Casos de Uso

01Generating AI-driven explanations and insights for machine operators
02Real-time monitoring and anomaly detection in industrial machinery and processes
03Integrating conversational AI capabilities into Human-Machine Interfaces (HMIs) for interactive data querying