HD Hyundai Robotics Cuts Robot Troubleshooting Time by 63 Percent Using Palantir AI Agents

HD Hyundai Robotics has reduced the average time needed to resolve problems in robot programs from 21.5 days to 7.9 days since deploying AI agents developed with Palantir Technologies in August. The South Korean company, a unit of the HD Hyundai shipbuilding group, presented the results at Palantir's Sovereignty Bootcamp in New York.

Chief Technology Officer Ahn Seong-hwan delivered the keynote. He outlined the firm's strategy for physical AI and its collaboration with Palantir. Previously, engineers manually inspected robot movements and repeatedly adjusted sequences, routes and speeds. The process was difficult to standardize and took weeks. AI agents now analyze existing programs and suggest improvements. Engineers then review and refine those suggestions using their knowledge of the factory floor.

The company expects the approach to cut robot programming costs by more than 50 percent. Palantir's platform helps identify new business opportunities while keeping core data under the customer's control. Palantir Chief Revenue Officer Ryan Taylor said the firms would connect HD Hyundai Robotics' robots with Palantir's AIP platform to create further opportunities.

Ahn attended the bootcamp, which focuses on strengthening AI-driven competitiveness while protecting corporate data. The partnership aims to change how engineers oversee industrial sites. Routine programming work shifts toward problem definition and implementation of AI-based solutions. Field engineers receive training to work with the agents rather than replace them.

HD Hyundai Robotics has applied the system to its own operations and plans to expand the scope of physical AI solutions. The reduction from 21.5 days to 7.9 days covers the period after the August pilot. The company has not released the exact number of cases or the full methodology behind the average, but it described the drop as measured against prior manual workflows on the same class of robot programs.

Physical AI refers to systems that combine machine learning with real-world machinery, sensors and control software. In this case the agents operate on existing robot code rather than generating entirely new programs from scratch. Engineers retain final authority over changes. The model is intended to accelerate iteration without removing human oversight of safety-critical industrial equipment.

Palantir has positioned its tools for use in environments where organizations want to retain sovereignty over data and models. HD Hyundai Robotics cited that feature as a reason for the partnership. The Korean firm operates in shipbuilding, construction and other heavy industries where robot deployment faces tight schedules and high costs for customization.

The results were announced on Sunday, 1 November. Korean outlets including Yonhap and the Korea Herald carried the figures and the keynote details. The companies plan to continue the collaboration and identify additional projects that can show measurable results in short periods.

The time reduction applies specifically to troubleshooting and program adjustment, not to the initial design of new robot cells. HD Hyundai Robotics has not stated whether the same percentage gain has appeared across all customer sites or only in its internal pilots. It has indicated that the cost-cutting expectation for programming follows from the shorter cycle times.

Industrial robotics has long required specialized programmers for each deployment. The shift described here moves part of that labor into an agent that proposes edits, with engineers acting as reviewers. Whether the 63 percent reduction holds as the system scales to more complex or multi-robot lines remains an open operational question for the firm and its customers.