If you’ve worked in technology long enough, you’ve seen this corpse before.
It lies in the boardroom, surrounded by frustrated executives and exhausted engineers. The deceased’s name? “GenAI Project 2024”. The cause of death? Tool Overdose, Purpose Bankruptcy.
As a Product Manager focused on AI, I have performed dozens of these autopsies. The pattern is frighteningly consistent. In 85% of cases (real Gartner statistic), the project didn’t fail because the AI hallucinated or because the model wasn’t sophisticated enough.
It failed because someone bought a titanium hammer with a laser sight and went looking for nails, screws, or anything they could hit.
The Inverted Cycle (The Recipe for Disaster)
The fundamental error is the inversion of value logic. It happens like this:
- Hype: The CEO reads about ChatGPT/Claude/Gemini on LinkedIn.
- Tool: The order comes down: “We need to use Generative AI in everything!”.
- Data: The team rushes to throw whatever data they have into the model.
- Pain (Delayed): Only at the end does someone ask: “Wait, what problem are we solving again?”.
The result? Chatbots that no one uses, report generators that hallucinate critical data, and an astronomical cloud bill. You built a brilliant solution for a problem that didn’t exist.
The Correct Cycle: The Centrato Framework
At Centrato, we have zero tolerance for “AI by the pound”. Value engineering requires discipline. The non-negotiable cycle for any successful project is:
1. The Pain
Forget AI. What is the bottleneck? Where does your team bleed efficiency?
- Wrong: “I want a chatbot.”
- Right: “My support team spends 4 hours a day answering the same 5 questions about order status.”
2. The Fuel (Data)
Do you have the data to solve this pain? Is it clean? Accessible?
- AI is not magic; it is statistics applied to data. If your data is garbage, your AI will be a very eloquent garbage generator.
3. The Tool
Only NOW do we talk about technology.
- Maybe you need an LLM. Maybe you need a simple
if/else. Maybe you need an Excel spreadsheet. - The good AI PM chooses the minimum necessary tool to solve the pain, not the most expensive one.
4. The Value (Measurement)
How will we know if it worked?
- Define business KPIs, not technology KPIs. Don’t measure “model accuracy”; measure “hours saved” or “increase in conversion”.
Survival Checklist
Before approving the next budget, do this quick diagnosis:
❌ NOT-TO-DO (Danger Signs)
- The project started with “Look what this tool does!”
- There is no business problem owner (only technical owners).
- The success metric is “launching the AI” (launching is not success, it is cost).
- You are trying to replace complex human judgment without “human-in-the-loop”.
✅ TO-DO (Health Signs)
- You can describe the problem in one sentence without using the word “AI”.
- You mapped the necessary data flow before choosing the model.
- There is a clear financial or operational KPI tied to the project.
- You started small (MVP) to validate the value hypothesis.
Conclusion
AI is the most transformative technology of our generation, but it does not forgive a lack of strategy. Don’t be the executive who builds the roof before the foundation.
Start with the pain. Respect the data. And only then, choose the tool.
Your budget (and your sanity) will thank you.
