AI Agents: Why They Fail Despite Heavy Investment, and How to Succeed
Session Overview
Despite growing adoption of AI Agents, more than 95% of enterprise projects fail to deliver ROI. This session will define how AI Agents differ from simple chatbots and analyze the core reasons behind failure—accuracy gaps, high costs and latency, and insufficient stability or operational maturity. Drawing from success cases, it will present strategies for building AI Agents that are accurate, efficient, and reliable.
Finally, the session will introduce a survival roadmap for enterprises—AI Orchestration through GPUOps, MLOps, and LLMOps—alongside VESSL AI’s vision for a Multi-LLM Token Factory and a Self-Evolving AI Agent System.
Speaker
.png)
Ahn Jae-man is the CEO of VESSL AI. After completing his undergraduate studies at KAIST, he led mobile game infrastructure at Devsisters and later served as a technology executive at medical AI startup AITRICS, building deep expertise in DevOps and machine learning. Today, he is recognized as one of Forbes’ “AI Founders to Watch in 2025” and serves as a board member of Korea’s Large-Scale AI Promotion Council, contributing actively to the global AI ecosystem.