Every time I see a roadmap filled with AI features, I think about the 100 cup holder problem.
Because this pattern shows up whenever product teams mistake feature requests for real user needs.
Imagine you surveyed drivers about what they want in a new car.
A common complaint might be:
“There aren’t enough cup holders.”
If a product team took that feedback literally, they might design a car with 100 cup holders. Technically, they solved the problem. But no one would actually want the car. Because what drivers really meant was something different.
They wanted convenience. Storage. A thoughtfully designed interior.
The cup holder complaint was just the visible symptom.
Customers Describe Frustrations, Not Solutions
One of the first lessons product leaders learn is this:
Customers are excellent at describing problems. They are terrible at designing products. Users tell you where something hurts.They rarely tell you what the right solution is. If you treat feedback like a feature backlog, the result is predictable.
Feature sprawl.
Complexity.
Products that feel like a pile of requests instead of a coherent system.
The Pontiac Aztek Lesson
The Pontiac Aztek is a classic example.
When it launched in 2001, it was supposed to be the ultimate lifestyle vehicle. It had camping attachments. Modular cargo systems. Dozens of niche features. Each idea made sense on its own.
Together they produced one of the most famously awkward cars ever built. The problem was not that the features were wrong. The problem was that the product had no clear job to do. It tried to satisfy too many interpreted requests instead of solving a coherent user problem.
Most bad products are not built by bad teams. They are built by committees.
Marketing adds a requirement.
Sales adds another.
Customer feedback introduces five more.
Eventually the roadmap becomes a compromise between dozens of partial signals. No one is wrong. But the product loses its center of gravity.
The AI Version of the 100 Cup Holder Problem
We’re seeing a modern version of this problem right now with AI. Teams ask customers what they want from AI. The answers sound like this:
“Add AI to summarize things.”
“Add AI to write emails.”
“Add AI to automate tasks.”
So the roadmap fills up with AI features.
AI chat.
AI assistant.
AI generator.
AI everywhere.
But often those features are the new cup holders. They respond to surface requests instead of solving the underlying job. The real question is rarely “Where can we add AI?”
The real question is: What decision, workflow, or outcome could AI actually improve?
A Quick Note on Feature Creep
If this pattern feels familiar, it is.
Product management has had a language for it for years.
The Kano Model describes how features fall into three categories: basic expectations, performance improvements, and delight features. Most feature requests fall into the first category. Adding more of them rarely increases satisfaction. It mostly increases complexity.
If you’re not familiar with the model, I wrote a short breakdown
The Real Job of Product Leadership
Great product leaders do something subtle. They listen to what customers say. Then they reinterpret it. They ask:
What job is the user trying to accomplish?
What friction are they actually experiencing?
What outcome are they chasing?
Only then do they design the solution.
Across multiple companies and platforms, the pattern is consistent.
Customers ask for features.
Great teams uncover systems.
Customers describe symptoms.
Great teams discover the real problem.
And the best products rarely look like the original request.
If you ever find yourself building the 100th cup holder, pause. You may be solving the wrong problem. Because great products are not built by implementing requests. They are built by understanding what users are actually trying to achieve.