A launch platform, run weekly rather than daily. That is the entire product decision. On a daily launchpad a founder gets a few hours near the top before the next cohort buries them; on a weekly one, a launch has a week to be found, and the leaderboard is worth checking on a Tuesday.
The revenue comes from the same asymmetry every launch platform has: getting seen is worth money to the person launching, and nothing to everyone else. Scheduled launches, featured placement, sponsor slots, a paid tier, and credits to relaunch something that has changed enough to deserve it.
The quieter half is the free tooling around it: an SEO scanner and an Open Graph checker. They cost nothing to run, they are useful on their own, and they bring in exactly the people who are about to launch something. A directory that only asks is a directory nobody returns to.
Python and Flask again, on the same shape as the other directories: SQLite while it was being built, Postgres once it was real, Tailwind on the front. Sold, and still running under its new owner.