LG Innotek Builds AI System to Predict Component Prices, Boosting Order Competitiveness

LG Innotek announced on the 21st that it has built an AI system for finding the optimal components to apply to new products. The system is called the AI Component Proposal System, and development took about 2 years. It has recently been fully deployed across all business units.

LG Innotek's main product lines include camera modules, semiconductor substrates and automotive components, and the number of components mounted in major products averages hundreds. In the existing quotation and product development process, information was dispersed across multiple internal systems.

As a result, the existing process took a significant amount of time to verify information and identify new components. LG Innotek applied AI to the component search process to improve this.

The system was designed to go beyond component recommendations and select the optimal component by considering price and performance at the same time. The goal is to improve the completeness of products proposed to customers and strengthen order competitiveness.

LG Innotek integrated data on about 2 million internal and external components, including capacitors and inductors, and built an AI-usable database by standardizing component names, specifications and units that differ by manufacturer. The company created a foundation that allows AI to use dispersed component information by standardizing and integrating it.

The system, called the AI Component Proposal System, searches the entire database for components with similar conditions based on user-input specifications and presents them. Its scope includes both existing purchased parts and external market data, expanding the search range and reducing dependence on employee experience and existing purchasing history.

Its core function is AI-based Reference Price calculation. The AI analyzes actual purchase histories and price trends for components with similar specifications to present the current appropriate price range, and LG Innotek said the Reference Price has a reliability of more than 96%. This makes it possible to estimate the price range for new components with no existing purchase history.

Component candidates with price competitiveness can be selected within 2 hours, after which actual quotes are checked from the shortlisted suppliers. This process improves the speed of selecting the optimal component.

Industrial components vary in price depending on purchase volume, contract terms and purchase timing, and have characteristics different from consumer goods with fixed list prices. For that reason, it previously required separate quote requests to multiple suppliers, along with market price research.

LG Innotek applied the AI Component Proposal System to that process. A LG Innotek official said that AI systems for component recommendation have recently been appearing one after another, and that implementing the Reference Price function for each component is the biggest differentiator from existing systems.

With the system in place, LG Innotek cut the time needed to calculate quotes for new products by more than 70%. LG Innotek said it can now use the reduced time to review customer requirements in greater detail and improve the completeness of product proposals, while also quickly searching for substitute components if a specific part is discontinued or supply is disrupted. As a result, the company can respond more flexibly to supply chain changes.

Based on pre-collected and standardized internal and external component data, the system performs component search and price calculation. The system won the customer satisfaction award at the 2026 LG Awards in April, and the recognition signified that its innovative achievements were acknowledged.

LG Innotek plans to further develop the system by applying Agentic AI. The goal of the additional development is for AI to autonomously search and analyze the latest component information and reflect the search and analysis results in the system. Kim Jun-seong, head of LG Innotek's Purchasing Center and an executive vice president, said the AI Component Proposal System is optimized for manufacturing companies and for businesses that handle a wide range of components, and that it is a meaningful innovation that transforms the existing approach. He also said the company will continue AX-based ways of working to deliver value that exceeds customer expectations.

LG Innotek is expanding AX by applying AI across work beyond component purchasing and across production as a whole. In major manufacturing processes, it introduced AI Raw Material Incoming Inspection, which cut the time needed to analyze the causes of material defects by up to 90%. It is also applying AI Vision Inspection, which performs AI-based finished-product appearance inspection and defect detection, to camera module production and semiconductor substrate production.

LG Innotek introduced AI and robots into manufacturing processes and production site operations, pursuing both time savings and manufacturing innovation. It applied AI Process Recipe to optimize camera module production process conditions, reducing the search time from 72 hours to within 6 hours. It is also pursuing the establishment of Dream Factory, an FC-BGA production base that applies AI and robots across production.

Recently, it also applied EXAONE Tabular, LG AI Research's industry-specific AI model, to manufacturing sites. This reduced the learning time for changed process conditions by about 85%. At the same time, group-wide AX collaboration is also expanding.

LG Innotek became the first LG affiliate to introduce its own AX certification exam for employees. More than half of its office workers have now earned the certification. Based on this, the company is accelerating the internalization of employees' AI utilization capabilities.

Source: TECHWORLD · Park Kyu-chan
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407180

References

• https://www.epnc.co.kr/news/articleView.html?idxno=407180