What happens to the enormous sums companies spend on AI strategy — and why does so much of it go to waste? talks to Frank Meltke, a practitioner with nearly three decades of implementation experience who works across three very different worlds: Europe, the United States, and South Korea.
Frank is the founder and CEO of contraco, the digital transformation consultancy he established in 1998.
Frank works across Europe, the United States, and South Korea, where he has lectured at Kyung Hee University and gained a rare comparative view of how cultures adopt technology.
Key TakeawaysStrategy cannot be outsourced. Only the management board can write a company's strategy — AI strategy included.
Three cultures, three failure modes. German boardrooms ask “whether” (will the workforce accept it, is it compliant); American ones demand P&L impact within two quarters; Korea asks what the technology measurably produces.
A use-case stack is a warning sign. 720 slides with 600 use cases and no data audit is how a railway company wasted millions on predictive maintenance that had no maintenance knowledge to learn from. If the data is bad or missing, no AI can change that. Data first, narrow scope second, vendor last.
Ask the practitioner's questions. What happens if we do NOT use AI in three months? What can we achieve in a realistic time with a realistic budget? Which KPI will tell us?
Never fire your customer-facing people first. The company that cancelled its call-center contract before defining how quality would be measured watched its net promoter score fall below zero
Korea's real advantage is education, not technology. Universities grade how you understood and solved the problem. Regulation was co-written with practitioners as a framework of responsibility, not a list of forbidden use cases.
Frank's three golden rules. Understand the problem and you can solve it with AI too. Don't be afraid to say “I don't understand what this AI means.” And don't punish failed projects — treat failure as progress, and the next project improves.
Agentic commerce is arriving faster than governance. Alipay processed over 120 million agent-driven transactions in a week; European payment providers have no infrastructure to allow or refuse an agent's purchase. KYC must become Know Your Agent — and the EU AI Act does not cover it.
Someone must own the non-human workforce. Bots and agents are part of the value chain — invisible colleagues in the office.
The token-cost shock is six months away. Departments use AI freely; the bill arrives monthly and uncontrolled — a licensing model no CFO has governed before. Watch your token economics now (and see EP78 on hedging compute costs).
Chapters:00:00 Intro: Why AI strategy spending leaves companies frustrated
03:10 Consultants can guide strategy, but boards must own it
06:59 How European, US, and Korean priorities differ
10:57 The 720-page railway strategy that lacked usable data
15:33 When AI customer service drives NPS below zero
18:57 South Korea’s cross-disciplinary approach to AI education
21:03 AI literacy versus prompt-focused fluency
24:53 Why AI use and cheating are framed differently
28:52 Three practical rules for AI governance
31:01 Agentic commerce and the need to know your agent
34:12 Who is responsible for non-human employees?
39:35 Token costs and the next three years of AI adoption
contraco — Digital Transformation Consulting
Frank Meltke — personal site & publications
Anastassia Lauterbach — LinkedIn
AI Snacks with Romy and Roby on Spotify
“Leading Through Disruption” newsletter
First Public Reading, Romy, Roby and the Secrets of Sleep (1/3)
First Public Reading, Romy, Roby and the Secrets of Sleep (2/3)
First Public Reading, Romy, Roby and the Secrets of Sleep (3/3)
Romy, Roby and the Secrets of Sleep (book)
