What if the slickest movie discovery app you’ve ever used lived entirely in your browser — no login, no cloud bill, just pure, instant personalization?
That’s Cut, a web app born from tutorial hell and sheer stubbornness. Its creator, an amateur dev tired of half-built projects, finally shipped a prototype that swipes movies like Tinder for films: right for like, left for nope, down to skip. Trailers, posters, synopses, ratings — all powered by TMDb’s generous API. And here’s the hook: it learns you on the fly, without phoning home to any server.
Look, OTT giants like Netflix shove recent blockbusters down your throat, polluted by family binge histories or their own catalogs. Cut? It ignores that noise. Pick a mood — say, ‘gritty thriller’ — and it filters ruthlessly. Or let the genre weights evolve as you swipe.
Why Build Cut When Netflix Already Swipes?
But don’t OTTs do this? Yeah, they claim to. Except their recs skew new, shared-account generic, and platform-locked. Cut sidesteps all that — it’s yours alone, cross-mood, even sprinkling in Korean indies or Hindi gems alongside Hollywood. No ads, no upsells. Just discovery.
The dev’s blunt: > “Yeah, they do but they tend to only show you more recent movies and there’s also a feature in my app to recommend movies based on your mood right now. And also, your Ott might get used by your friends, your family etc. and that doesn’t allow for personalized selection and also the Otts only recommend movies from their catalog.”
Spot on. Corporate rec engines chase retention metrics; this one’s chasing your taste.
Infinite scroll feeds demand smarts. Cut’s algorithm starts naive — every genre (23 in TMDb) at 100 weight. Like an action flick? Bump action up. Hate romance? Slash it down. RNG picks weighted genres, fetches popular 7+ rated movies from TMDb. Boom: more of what you love, with discovery serendipity. The dev admits surprise: “I honestly thought wouldn’t work very well. But to my surprise it genuinely giving picks I liked.”
How Does Cut’s Genre Algo Actually Learn You?
Simple, right? Deceptively so. TMDb’s vast catalog — thousands per genre — lets weights shine. No ML black box, just arithmetic and probability. You’ve seen echoes in Pandora’s Music Genome, early 2000s radio that weighted traits sans servers. Cut revives that indie spirit in a post-AI world, proving heuristics crush hype for solos.
Storage? LocalStorage. User’s genre prefs, seen movies (no repeats), watchlist — all persist till you nuke cache. No Firebase friction, no SQL slog, no auth walls. For a prototype eyeing zero users? Perfect. Critics howl ‘not scalable!’ Sure. But scaling starts with shipping. This dev nails the indie mantra: MVP in browser, iterate if traction hits.
Here’s my unique take — Cut signals a stealth architectural shift. We’re bloated on serverless cults, but local-first apps (think Dexie.js wrappers later) reclaim control. Remember Flickr’s early photo streams? Pure client-side magic before VC bloat. Cut’s localStorage bet predicts a wave: privacy-first recs as cookies crumble and GDPR bites.
UI tease: Keyboard shortcuts (‘w’ for watchlist), provider lists per film. Mood picker toggles feeds temporarily. Prototype skips accounts — smart for momentum. Next devlog promises watchlist deep-dive, polish.
Skeptical? Test the logic. Start weights even. Swipe 10 actions likes: action dominates, but RNG sneaks adventures. Dislike animations: they fade. Minimum 7 rating + popularity filters junk. Local langs via TMDb params. It’s not Recommender Systems 101; it’s scrappy effective.
Can Local Storage Power Real Movie Discovery Apps?
Absolutely — for now. Persists across sessions, device-local. Multi-device? Export JSON later. Users clear cache? Reset and relearn — low friction. Vs. Firebase: no vendor lock, no bills. Dev candid: “I also didn’t have the budget to make or rent servers or the time to learn SQL.”
Bold prediction: If Cut hooks 100 users, it’ll spawn clones. Add PWAs for offline, Supabase for sync. But root it local-first? That’s the genius. Big Tech forgot: users crave ownership.
Problems faced? API throttling (40/min ample), no dupes via storage, mood-state persistence. All crushed sans frameworks bloat — vanilla JS? Implied, keeping it lean.
This devlog thrills because it’s human. ‘Tutorial hell’ scars, determination wins. Suggestions pour in comments; iteration beckons.
Cut exposes OTT underbelly: personalization as myth when shared. Your solo browser app? Truth.
🧬 Related Insights
- Read more: Headlamp’s 2025 Surge: Kubernetes UI Finally Feels Native
- Read more: Tailscale’s Quiet Pivot: Tailnet to Identity-Driven Platform with TSIDP and Aperture
Frequently Asked Questions
What is the Cut movie app?
Cut’s a web-based movie discovery tool using TMDb: swipe likes/dislikes, mood filters, genre-learning algo, all in localStorage.
How does Cut’s recommendation algorithm work?
Genre weights start equal, adjust on swipes; RNG picks weighted genres for 7+ rated popular movies, avoiding repeats.
Why use local storage for Cut instead of a database?
Keeps it prototype-simple, no auth/servers needed; persists user prefs till cache clear, scales to real users later.
