[Edaily Reporter Yun Junghoon ] Ibex, a company specializing in industrial physical AI, has concluded its first annual forum, where it shared implementation strategies for industrial sites as they expand from enterprise AI to physical AI.

AIVEX announced on the 12th that it held its annual AI forum, “AIVEX FRONTIER 2026,” on the 11th at The Westin Seoul Parnas in Gangnam-gu, Seoul. Approximately 300 attendees, including executives from major manufacturing and IT companies, AI experts, and investors, participated in the event to discuss topics such as △the current state of enterprise AI, △business expansion, and △bridging the gap between goals and reality.

In his keynote speech, AIVEX CEO Seong Min-soo identified the fragmented structure—where optics, modeling, mechanics, control, and operations are handled separately by different companies—as the reason why proof-of-concept (PoC) projects for physical AI on the manufacturing floor fail to translate into actual mass production.

To overcome this challenge, iBex has directly developed five technology layers—ranging from sensing (L1) to recognition (L2), mechanics (L3), control (L4), and learning/operation (L5)—and supplies them under a single point of responsibility. The vision foundation model applied to mass production lines inspects thousands of product varieties using a single model, reducing the amount of training data to one-thirty-sixth of the previous requirement. It also features a “closed-loop” structure that automatically retrains the model using on-site inspection data, enabling overnight updates without interrupting production lines.

Presentations followed that addressed monetization and infrastructure strategies for enterprise AI transformation (AX). Kim Ji-hoon, Managing Partner at Boston Consulting Group (BCG), stated, “The revenue improvement effect for AI leaders is five times that of latecomers,” and emphasized the need to build “Enterprise IQ” by structuring a company’s unique business logic.

Min Se-hoon, Head of Commercial Operations for APAC at Palantir Technologies, called for profit and loss (P&L) improvements through an ontology-based “Enterprise OS,” while Ahn Jae-man, CEO of Vessel AI, presented an operational strategy for a hybrid “AI Factory” tailored to an era of surging token consumption.

Solutions for the commercialization of physical AI and robotics were also shared. Park Ji-hoon, Group Leader at iVex AI-ROBOTICS, introduced “hybrid control”—which combines teaching, vision-guided, and Visual Language Action (VLA) models—and the integrated platform “AIVot” for production sites requiring a success rate of over 99%.

Noh Beom-jun, Executive Vice President of Business Development for AI Home Solutions at LGELECTRONICS(066570), presented the “Zero Labor Home,” designed to reduce the burden of household chores, along with examples of space-based physical AI. Professor Kim Geon-hee of Seoul National University discussed the “World Action Model,” while Choi Hoon, Head of the Enterprise Business Division at Upstage, addressed field-centric AI adoption strategies.

The final speaker, Professor Dennis Hong of UCLA’s Department of Mechanical and Aerospace Engineering (Director of RoMeLa), emphasized the need for robot designs optimized for specific tasks, stating, “Humanoids that mimic human form face limitations in structural complexity, energy efficiency, and cost.”

CEO Seong Min-soo said, “We will continue to enhance the applicability and stability of AI in industrial settings by providing turnkey solutions that integrate inspection, robotics, and MLOps into a closed-loop structure and by internalizing the underlying technology layers.”