DEV Community

Cover image for I built TrendingTide: Hacker News trends with AI key takeaways
Digiwares
Digiwares

Posted on

I built TrendingTide: Hacker News trends with AI key takeaways

I built TrendingTide because I kept finding useful Hacker News threads, opening too many tabs, and then losing the good discussions later.

TrendingTide is a small tool that helps you scan live Hacker News trends, summarize noisy comment threads into key takeaways, and save useful discussions without signing up.

👉 https://trendingtide.xyz

What it does

TrendingTide currently supports:

  • live Hacker News trend feed
  • Tide Score for ranking active discussions
  • topic filters for AI, startups, developer tools, security, Show HN, and Ask HN
  • AI-generated KTA summaries
  • saved threads
  • local history
  • KTA from saved/history pages
  • no signup
  • mobile-friendly terminal-style UI

KTA means “key takeaways.”

When you click KTA, TrendingTide reads the public Hacker News discussion and summarizes:

  • key takeaways
  • why it matters
  • possible opportunity

Why I built it

Hacker News has a lot of signal, but the best value is often buried inside long comment threads.

Sometimes a thread is not just “news.” It contains:

  • founder pain points
  • developer tool complaints
  • AI workflow debates
  • security concerns
  • pricing lessons
  • product feedback
  • early technical signals

I wanted a way to quickly understand whether a thread was worth reading deeply.

How it works

TrendingTide uses:

  • Next.js
  • Hacker News API
  • Claude for summaries
  • Upstash Redis for caching, rate limiting, and locks
  • Vercel for hosting and analytics

The KTA flow is intentionally on-demand.

When someone clicks KTA:

  1. The server checks Redis for a cached summary.
  2. If it exists, it returns the cached version.
  3. If not, it collects public HN comments and asks Claude to summarize.
  4. The result is stored in Redis.
  5. Future users get the cached KTA.

There is also a Redis lock per story so if 100 users click KTA on the same uncached thread at the same time, they do not trigger 100 AI calls.

Local-first saves

Saved threads are stored in browser localStorage.

That means:

  • no account required
  • no signup wall
  • no database needed for personal saves
  • users can save useful discussions immediately

If you generate a KTA and then save the thread, the KTA is saved locally too.

What is next

Right now TrendingTide is focused on Hacker News.

The next phase is to add a real historical data layer with Supabase:

  • persistent story history
  • trend pages like /trend/mcp or /trend/ai-agents
  • weekly intelligence pages
  • trend velocity
  • topic timelines
  • founder opportunity extraction

The bigger idea is not just “summarize Hacker News.”

It is:

Extract useful intelligence from technical internet discussions.

Try it

You can try it here:

https://trendingtide.xyz

Feedback is welcome, especially on:

  • KTA quality
  • topic filters
  • what sources to add next
  • whether saved/history workflows feel useful

Top comments (0)