LG Advances Factory AI Integration with EXAONE for Enhanced Production Insights

Key Takeaways

LG AI Research unveiled two AI models, EXAONE Tabular and EXAONE Omni-Inspect, aimed at improving manufacturing processes.
EXAONE Tabular utilizes in-context learning to analyze production data and adapt to changes with minimal data input, significantly reducing retraining time.
EXAONE Omni-Inspect enhances visual inspection processes by detecting defects in products while adapting to changes in manufacturing conditions.

Introduction of AI Models for Manufacturing

LG AI Research recently introduced two innovative AI foundation models, EXAONE Tabular and EXAONE Omni-Inspect, presented at the LG AI Talk Concert 2026 in Seoul. These models target manufacturing capabilities, focusing on production-data analysis and automated visual inspection. They are designed to adapt to changing production conditions while minimizing the need for additional training.

Features of EXAONE Tabular

EXAONE Tabular specializes in the analysis of structured data, such as manufacturing processes and product quality. It identifies relationships within numerical data and uses them for predictive insights about production. Notably, this model can make forecasts with smaller datasets, an approach that decreases data collection efforts when production environments change.

In contrast to traditional methods, EXAONE Tabular employs in-context learning, enabling predictions utilizing prior examples without extensive retraining or dataset-specific adjustments. This capability reportedly led to an 85% reduction in the time needed to adapt models to production changes. The model demonstrates efficiency with a relatively modest size, containing only around 21 million parameters while producing performance metrics comparable to larger models like Google’s TabFM.

Enhancements via EXAONE Omni-Inspect

EXAONE Omni-Inspect addresses the need for visual quality inspections, using camera analysis to detect defects in both components and final products. A key aspect of this model is its ability to continue functioning effectively even when new products alter manufacturing appearances, thus eliminating the necessity for frequent retraining.

The company is also advancing a vision inspection agent, which is anticipated to automate data sampling, labelling, and model training, further streamlining operational adaptations.

LG’s Commitment to Industrial AI

Since its inception in December 2020, LG AI Research has tackled over 100 industrial challenges, including battery life prediction and defective-product detection. The focus remains not only on creating robust AI models but also on addressing long-standing industrial issues that require complex solutions.

Moreover, LG’s EXAONE On-Premise system launched in July 2025 aims to provide enterprises with a secure full-stack solution, ensuring sensitive data remains within corporate environments. However, it remains unclear if the new models will operate via this system or through other infrastructures at the factory floor level.

Future Developments in Robotics

LG is also making strides in the robotics sector, developing robot foundation models that will allow automated factories to enhance their operational capabilities. These models are designed to enable robots to assess conditions, make decisions, and perform actions safely, steering toward a fully autonomous manufacturing ecosystem that interlinks various systems across production environments.

As such advancements continue, LG’s vision of integrating AI-driven manufacturing solutions is clearly on the horizon, paving the way for smarter, more responsive factories equipped to handle evolving production demands.

The content above is a summary. For more details, see the source article.