LG Electronics has reportedly built an artificial intelligence (AI) system that can cut the parts review process during new appliance development—from what used to take several days—to about 30 minutes. The company is conducting a proof of concept (PoC) and aims to roll it out in-house within the year.
Since taking office as chief executive officer (CEO) early this year, Lyu Jae-cheol, president of LG Electronics, has consistently emphasized improving work efficiency through an "AI transformation" (AX). As Lyu's management philosophy is applied to the actual research and development (R&D) environment, it is changing the "way of working."
According to reporting compiled on the 30th by ChosunBiz, LG Electronics has developed an AI-based parts search agent and is currently running a PoC. To build this system, LG Electronics converted design materials accumulated over decades into "AI-ready" data that AI can learn, analyze, and reason over. On this basis, it built the parts search AI agent "Part-Riever." Part-Riever combines "part," meaning component, with "retrieve," meaning to get back.
A single large appliance such as a refrigerator, washing machine, or dryer contains more than 1,000 parts. Engineers at LG Electronics had to search, compare, and review data on hundreds of thousands of parts every time they developed a new product.
There were multiple constraints on accessing design data. For security and other reasons, it was stored dispersed across multiple systems. Most of it also existed in unstructured forms such as 2D drawings or 3D shape data, which meant engineers spent enormous time finding the information they needed. LG Electronics employees had relied on related departments for parts searches or handled them based on past development histories.
Through Part-Riever, LG Electronics has shifted the conventional parts review method—dependent on personal experience and past records—to an AI-based data analysis framework. By processing design data into a form AI can understand and use, the company greatly improved accessibility. Even new engineers can now use the accumulated design assets to review parts and make decisions without consulting veteran engineers.
The core of Part-Riever is natural language search. When engineers enter conditions as they would normally speak, the AI finds suitable parts from existing design assets. For example, if they enter, "Find a heater component made of stainless steel, similar to the existing model, with output of at least 1,000 watts (W)," the AI goes beyond simple keyword search to deliver results by comprehensively analyzing diverse data such as 2D drawings, 3D shapes, and technical documents. According to LG Electronics, even if part names or specifications differ, it typically takes only 1–2 minutes to find and propose the optimal similar part.
To realize these functions, LG Electronics extracted and structured key information—such as specifications, materials, and drawing tolerances, which are production conditions—from 2D drawings. It converted 3D shape data into direction-sensitive vectors. This created a foundation for AI to judge similarities between parts on its own.
After applying Part-Riever to certain tasks, LG Electronics cut processes that previously took several days to about 30 minutes. In the recent development of a new electric cooktop, Part-Riever analyzed some 550 candidate parts and successfully proposed suitable similar parts. The number of parts requiring new development also fell by about 25%. By standardizing verified parts, the company simultaneously improved development speed, production efficiency, and design quality.
LG Electronics plans to use Part-Riever not only for new product development but also to drive quality improvements. Reusing verified parts across multiple product lines can reduce quality variance and defect risks. As the range of identical-part application widens, the company also expects faster parts sourcing and after-sales service (A/S) response in the service phase. It also aims to strengthen competitiveness across the supply chain by boosting suppliers' parts production and operational efficiency.
LG Electronics plans to complete the ongoing PoC, make any necessary refinements, and deploy Part-Riever in-house within the year. Going further, it will convert data across manufacturing and development—starting with design data—into assets usable by AI, and accelerate AX-centered manufacturing innovation.
The development of Part-Riever at LG Electronics is said to have proceeded under strong backing from Lyu. Since taking office, Lyu has emphasized that, through AX, it is important to redefine the "way of working," raise work efficiency, and focus capabilities on high-value matters. At the first press conference after taking office in Jan., Lyu said, "With the advent of AI, the way we work has changed," adding, "By leveraging AI, we will build 'quality, cost, delivery' (Quality·Cost·Delivery) competitiveness so we can win in the speed game."
An LG Electronics official said, "The assetization of design data and the development of AI agents is the process of converting personal experience and know-how into assets that the entire organization can use," adding, "We will continue to drive AI-based manufacturing innovation to improve development speed and quality and lead customer experience innovation."