'Agentic AI Training' – Instructor John Kim / Korea Expressway Corporation

Today I delivered an AI training session in Damyang, at the Korea Expressway Corporation Gwangju-Jeonnam Regional Headquarters. I never expected public-sector employees’ questions about using AI to be this specific. Questions poured in about AI agents like Claude Cowork and OpenClaw, and the four real cases that came up on site were impressive.

Case 1 — Automating management of expressway toll arrears. Picking out customers with arrears of KRW 1 million (≈ US$750) or more by hand in Excel, calculating reductions, and sending notices: today, people do it one by one. The AI agent scans the Excel files in a folder → filters targets over the threshold → applies a 30% reduction → and even sends emails and text messages automatically. Someone even asked, "Can AI make the phone calls too?" (It can )

 Case 2 — A data pipeline to decide on rest area expansion. The idea: use a single camera to capture vehicle entry frequency, vehicle type, vehicle length, and even weight from tire condition, then decide whether to expand a rest area. They were amazed when I shared the Input → Process → Output frame. We designed a workflow that organizes the raw data in Excel and then generates the report automatically.

 Case 3 — Turning 2D characters into photorealistic 3D shorts. The PR team at Korea Expressway Corporation wanted to turn its five 2D characters into photorealistic 3D versions and make talking short-form videos. Separate the characters → render them photorealistic (Banana) → build character sheets → test video models (Runway, Kling, Seedance, etc.). The key: "once you find the workflow, the next step is standardizing it."

  Case 4 — Automating facility inspections. Streetlights and structures are filmed with a dashcam → a person reviews the footage by eye to decide whether to replace them. Use that footage as AI training data to classify conditions automatically from video, producing the raw data and the report in a single workflow.

Automating work with AI raises efficiency. But when that efficiency starts to replace "my reason for being," what new value should we create? Think public institutions are slow to adopt AI? Seeing today’s session would change your mind. The issue isn’t the technology; it’s the quality of the questions. Today the Expressway Corporation team asked excellent ones.