Sentinel Devices offers OTAware, a plug‑and‑play, on‑premises module that connects directly to PLCs to collect machine data, run local machine‑learning for anomaly detection, and issue simple yes/no alerts without cloud connectivity. The solution provides offline data storage, bandwidth‑efficient alerts, and a drag‑and‑drop interface for operators to train models, reducing downtime and cybersecurity risk for industrial manufacturers.
Funding
Funding not disclosed
Founders
Product
Problem
Industrial facilities often struggle with complex, costly setups for equipment monitoring that require extensive networking, cloud infrastructure, and ongoing cybersecurity management, leading to high downtime and security risks.
Solution
OTAware is an end‑to‑end, plug‑and‑play platform that consolidates equipment monitoring, data storage, and cybersecurity into a single on‑premises module for each digital controller. The system continuously collects low‑level machine data, applies local machine‑learning models to detect anomalies, and generates simple yes/no alerts, eliminating the need for cloud connectivity. By keeping all processing and data within the facility, OTAware reduces the attack surface and removes reliance on external updates or third‑party servers. Automated anomaly detection provides 24/7/365 oversight, while a drag‑and‑drop interface lets operators train the model on normal operation periods, preserving knowledge across workforce changes. The platform stores billions of data points locally, enabling long‑term trend analysis and pinpointing the most likely fault signals without streaming large data volumes.
Target Audience
Primary customers are manufacturers and operators of industrial plants—such as chemical, power, and beverage production facilities—that rely on PLC‑based control systems and need secure, low‑maintenance equipment monitoring.
Features
- Plug‑and‑play module that connects directly to PLCs or other digital controllers, requiring same‑day installation and no retraining
- Fully offline, air‑gapped processing and storage to eliminate cloud‑related cybersecurity risks
- Local machine‑learning engine that autonomously learns normal operation and flags deviations with yes/no alerts
- “Blame Support” feature that ranks sensor signals by out‑of‑line behavior, guiding rapid fault diagnosis
- Data profiling capability that retains historical data (up to 30 billion points) and supports seamless transition when equipment is reconfigured or upgraded
- Drag‑and‑drop training interface allowing operators to label normal operation periods without specialized data‑science expertise
- Bandwidth‑efficient communication that transmits only essential alert signals, reducing network load and cost