So I started reading Reddit and HN threads looking for people with real pain points. It worked, but I was spending hours a week just reading threads and not building anything.
After a few weeks, I built a scraper. Which evolved into a full extraction pipeline through Claude. The prompt took some time to get right but now it's pulling out the specific pain point, a target audience, a direct quote from the source, what people are doing instead, and a rough read on competition and effort.
One from this week: a backend engineer wrote "even with a decent CI/CD pipeline, the first 10-15 min after a production deploy are strangely manual." Small teams with no Datadog budget just watching dashboards by hand after every push. This gap has existed for a while but it's getting worse with AI increasing deploy frequency.
Feed is free to browse, filterable by industry and effort. PMFounder.com
Happy to get into the extraction prompt design or pipeline architecture if that's interesting.