ReplayAutonomous app testing for teams who ship at AI-speed
Promoted
Trending AI
Global tech trends, fast-picked by AI.
5 followers
Global tech trends, fast-picked by AI.
5 followers
Trending AI pulls GitHub Trending, Hacker News and Product Hunt into one open-source app. An LLM scores the noise and picks a few items a day, each with a short structured take: why it matters, what it's for, and the alternatives. English or Chinese, your choice. Built entirely in Kotlin Multiplatform, MIT licensed. Android is shipping on Play and F-Droid; iOS compiles and runs but isn't polished yet.
Hi Product Hunt!
Fun fact: Trending AI has been reading the PH daily leaderboard since day one — so today it's on the other side of the feed. :)
I built it to fix my own mornings: skimming GitHub Trending, Hacker News and Product Hunt to find what's actually worth reading. The app pulls from all three (PH via the official API), and an LLM picks a handful of items a day — each with a short structured take: why it matters, what it's for, the alternatives. English or Chinese, your choice.
Honest status: Android is shipping (Play, F-Droid, GitHub). iOS compiles and runs in the simulator but isn't polished yet — the whole client is one Kotlin Multiplatform codebase with Compose Multiplatform UI, MIT licensed.
Would love feedback — especially from fellow PH lurkers on whether the daily picks match what you'd have clicked anyway.
Repo: https://github.com/HarlonWang/TrendingAI
The kmp stack is nice and it actually feels lighter than my usual feed reader, though the summaries can be a bit surface-level for niche repos.
@nurglsatolb7dx Thanks — "lighter than my usual feed reader" is exactly what I was aiming for. And fair point on niche repos: when there's little public context, the model used to lean on metadata alone and the take got generic. I recently shipped a change that feeds the actual README/article body into the summarizer, so newer summaries should have more substance. If you hit a repo where it still feels shallow, I'd love the link — those make the best test cases.
Cool approach, love that it's Kotlin Multiplatform and open source. One thing that would help me a lot: let me star or save the picks I actually click through on, so the LLM learns what kind of stuff I care about and the daily shortlist gets more tailored over time.
@narinbily7re Thanks! And honestly that's the feature I want most myself. Where it stands today: the daily picks are generated server-side and are the same for everyone, so a star/save button is the natural first step — then using those signals to re-rank the shared shortlist on-device, rather than running per-user LLM passes. If you're up for it, open an issue on the repo with how you'd expect it to behave — it's all MIT and I'd love to spec it out in the open.