Mexin Labs provides rugged edge computing appliances that run AI inference locally on industrial sites, integrating GPU/TPU and FPGA accelerators with a modular software stack for computer‑vision and sensor‑fusion workloads. The devices connect to legacy control systems via Ethernet, CAN, Modbus and other interfaces, processing data on‑premise to meet low‑latency, privacy and compliance requirements.
Funding
Funding not disclosed
Founders
Product
Problem
Industrial operations often require real‑time AI inference for tasks such as predictive maintenance, visual quality inspection, or asset monitoring, but limited or unreliable network connectivity, high latency, and strict data‑privacy regulations make cloud‑based solutions impractical.
Solution
Mexin Labs delivers purpose‑built edge computing units that execute AI models directly on the shop floor, field site, or logistics hub. The hardware is ruggedized for harsh environments and integrates GPU/TPU accelerators with FPGA‑based inference pipelines, enabling millisecond‑level response times without reliance on external networks. A modular software stack provides pre‑configured computer‑vision and sensor‑fusion workloads—e.g., defect detection, equipment fatigue analysis, and precision farming analytics—while allowing customers to upload custom models via a secure SDK. All data is processed and stored locally, preserving confidentiality and meeting industry‑specific compliance standards. The solution is packaged through a three‑phase rollout: feasibility assessment, custom hardware/software integration, and proactive post‑deployment support, ensuring minimal disruption to existing PLC, SCADA, or ERP systems.
Target Audience
Primary customers are manufacturers, retail chains, agribusinesses, construction firms, and logistics operators that need on‑site AI for process optimization, quality control, or safety monitoring.
Features
- Rugged edge appliance with integrated GPU/TPU and FPGA for on‑device AI inference, supporting TensorRT, ONNX, and OpenVINO models
- Plug‑and‑play I/O suite (Ethernet, CAN, Modbus, USB, optional 5G) for seamless connection to legacy sensors and control systems
- Pre‑installed computer‑vision pipelines for defect detection, inventory verification, and crop health assessment, customizable via a Python/C++ SDK
- Real‑time digital‑twin synchronization engine that streams state vectors to local dashboards without cloud transit
- End‑to‑end encryption (TLS 1.3) and secure boot, ensuring data remains on‑premise and tamper‑proof
- OTA firmware and AI model updates managed through a centralized device‑management console
- Low‑power design (≤30 W) enabling operation on battery or solar‑powered edge sites
- Integrated health monitoring and predictive diagnostics for the edge unit itself, reducing downtime