Manufacturing AI Expertise Drives US$3M Defense R&D

SURROMIND Begins Developing an AI Agent-Based Vehicle Predictive Maintenance System to Enter the Defense AI Market

Detecting vehicle anomalies early to support maintenance, with a target preventive maintenance transition rate of at least 70%

SEOUL, October 1, 2026 — SURROMIND, a company specializing in manufacturing AI led by CEO Hayoung Jang, has begun R&D on an AI agent-based vehicle predictive maintenance system. The project will develop core technologies to shift maintenance of medium standard military vehicles from post-failure responses to AI-based predictive and preventive maintenance. Conducted under a dedicated technology development support program for Defense Innovation 100 companies, the project has a total budget of approximately US$3 million (KRW 4 billion, based on an exchange rate of KRW 1,350 per US$1) and will run for 28 months, from September 2026 through December 2028.

Conventional vehicle maintenance has focused on diagnosing faults and making repairs after a failure occurs. This project will develop a system that detects changes in vehicle condition early, allowing maintenance to be planned when signs of an anomaly first emerge. Its central aim is to shift from reactive to preventive maintenance by preparing the necessary inspections and corrective actions before a failure occurs.

SURROMIND will develop four AI agents, each responsible for analysis, diagnosis, prediction or maintenance guidance. The analysis agent checks vehicle condition and identifies anomalies. The diagnosis agent identifies fault types and causes, while the prediction agent estimates remaining useful life and maintenance urgency. Based on these results, the maintenance guidance agent provides maintenance personnel with inspection and corrective-action procedures, along with the parts and tools they need.

▲ How the AI agent-based vehicle predictive maintenance system works

Diagnostics will cover three vehicle systems: engine, cooling and electrical. After maintenance, field findings and vehicle condition data will be fed back into AI training and validation to continuously improve the system. Through this R&D project, SURROMIND aims to achieve a preventive maintenance transition rate of at least 70% and reduce on-site diagnosis time by at least 50%.

▲ Targeted outcomes

With this R&D project, SURROMIND is expanding its business from manufacturing AI into defense AI. The company plans to extend the technological foundation it has built as a manufacturing AI specialist to defense applications and step up its entry into the defense AI market with its AI agent technology.

“Even as maintenance personnel numbers decline, military equipment availability and combat readiness must be maintained,” said Byunghee Kim, CTO of SURROMIND. “We will apply the AI agent technology we have built up in manufacturing to military maintenance, developing a maintenance AI agent system that personnel can use in the field, from anomaly diagnosis through corrective action.”

Website: https://surromind.ai/