I stopped brainstorming startup ideas — this worked better for me

For a long time, my “idea generation” process was:
• notes
• tweets
• random thoughts
• endless lists

It felt productive, but it wasn’t grounding.

Lately, I’ve been experimenting with a different approach:
starting from real user frustration, not ideas.

Instead of asking “what should I build?”, I look at:
• what people are repeatedly confused about
• where decisions stall
• what questions never get a clear answer

What surprised me most is how repeatable these problems are across different communities.

I’m now building a small tool (Problem Miner ) around this approach because it helped me personally stop guessing and start validating with more confidence.
scans real conversations from communities (like Reddit), then use a small pipeline to:
1. detect whether someone is expressing a real frustration (not advice or promotion),
2. summarize the core problem in plain language,
3. tag it with context (who it affects, domain, severity, willingness to pay),
4. store it so patterns can be spotted across many similar complaints.(clustering)

The goal isn’t to generate ideas — it’s to surface repeatable pain people already care about.

Curious how others here source ideas:
• do you start from problems?
• or do you refine ideas first and validate later?