Why Do 95% of GenAI Projects Fail? The Fatal Traps of Illusion
Session Overview
Across the globe, countless enterprises are pursuing Generative AI projects, yet only 5% have translated into actual revenue growth. In other words, 95% of projects fail to deliver the expected results. The reason lies not in the technology itself, but in the absence of an “execution DNA”. This session, based on real-world experience building AI Agents for financial institutions and retail companies, will highlight the common patterns that cause Generative AI projects to collapse and present practical strategies to overcome them.
The talk will focus on four key pillars:
- Reviewing business objectives and ROI realistically – start from business, not technology
- Clear understanding of LLM strengths and limitations – balancing imagination with risk management
- Tool use and data readiness – how unprepared data drives failure
- Barriers of cultural transformation – organizational resistance and execution limits
Speaker
Han Seon-ho has extensive experience across industries including finance, healthcare, smart cities, higher education, and the public sector during his tenure at IBM. He previously served as Head of Watson & Cloud Business Unit and Head of Data & AI, leading AI-Data-Cloud–driven digital transformation initiatives over the past decade. Currently, he serves a dual role as Head of Financial AI at Bespin Global and CEO of Cognet9, an AI-focused MSP. In these capacities, he helps financial and retail enterprises achieve AI-driven business transformation through Scaling AI, Vertical AI innovation, and AI Service-as-Software, accelerating their journey toward AX