Lesson 03: Find the Problem First, Then Build

The AI Trap

AI makes building things absurdly easy. What used to take a week can now take an hour.

That creates a dangerous habit:

Building things because you can, not because anyone needs them.

The wrong question is: "What else can I build with AI?"

The right question is: "What do users actually need?"

Those two questions lead to completely different outcomes.

The First Two Questions for Any Product

Who is this for? Be specific. Not "people interested in Chinese culture." Name the actual person with the actual situation.

What problem does it solve? The answer should describe a real frustration, not a feature.

For example:

Weak: "I built a Chinese name generator." Strong: "A non-Chinese speaker wants a Chinese name that doesn't sound awkward to native speakers and actually means something good."

The second version describes a real person with a real problem. That's a product idea. The first version is just a description of a feature.

Need Signals Have Different Strengths

WeakestI think people would like this.
WeakI found a Reddit thread complaining about this.
MediumPeople are searching for this on Google.
StrongSimilar tools already exist and have real users.
StrongerUsers have directly asked for this.
StrongestSomeone paid money for something like this.

Payment is not the only signal, but it is the strongest one. It removes all the guessing.

Before You Build Anything, Fill This In

User:Who exactly is this for?
Problem:What frustration does this solve?
Alternative:What do they do today without this?
My solution:What does this do differently or better?
Why better:Why would they choose this over the alternative?
Proof of work:How will I know if it's actually working?

If I cannot fill this in before I start, I wait. The blank fields are the real work, not the code.

"Don't build because AI can. Build because users need it."

← Lesson 02 Lesson 04: Build → Measure → Decide →

Discuss this lesson on GitHub →