I built a free tool to find the exact communities where your target customers hang out

The distribution problem that kills most indie projects isn't the product — it's not knowing where your people are.

I kept finishing builds and then spending weeks guessing which subreddits, Discord servers, or forums to post in. Getting banned from one, ignored in another. All while the communities where my exact customer was already talking about the problem I solve were right there.

So I built NicheRadar.

What it does: You describe your product and your ideal customer in two sentences. It returns 8 real communities with:

• Live member count (verified via Reddit API for subreddits)
• Activity level (high/medium/low)
• The 3 pain points members are actively discussing
• A first post idea that passes the "product name deletion test" — must be genuinely useful to the community even if you removed any mention of your product
• Self-promotion rules so you don't get banned on arrival

The stack: Next.js + Claude Haiku 4.5. Each scan costs ~$0.006 in API calls.

What I learned building it: The hardest part wasn't the code — it was getting the AI to NOT hallucinate communities. I had to add explicit anti-hallucination rules: only return communities you're certain exist, verify member counts are realistic, no invented subreddits. Still not perfect (non-Reddit sources use model estimates, not live data), but useful enough that I'm using it for my own distribution.

It's free right now. I'm collecting feedback before deciding what to charge.

https://radar-v2-theta.vercel.app

Question for IH: How did you find your first 10 communities for distribution? Did you do it manually or have a system?