A movie recommender that learns your taste on your phone — and keeps it there. No accounts. No posters. No infinite scroll. Just fifteen well-chosen films, every morning.
Most film apps are billboards in disguise — designed to sell posters, sell engagement, sell you. WhatNext is the opposite: a quiet, on-device reader that learns what you love and delivers a short list each morning.
No sign-up, no email, no cloud. The app opens, you start. Your taste profile is a single file on your device — back it up if you like, or don't.
Export it, delete it, restore it on a new device. Your ratings and taste profile are a portable file — yours to keep or wipe entirely. No company holds a copy.
Rate a few films you love and the model learns your taste — on this device, in milliseconds, and never anywhere else. No algorithm serving engagement. Just films that fit.
Setup takes under three minutes and never leaves your phone. The recommender runs locally on a small taste vector — no servers, no telemetry.
Tap ten films you've enjoyed from a curated grid of two hundred. Or skip the grid for a tag-based fallback if your watch list is shorter.
One card at a time. Mark each as Enjoyed it, It was okay, or Not for me. The model calibrates against your replies.
Each day, fifteen films are queued for you — synopsis, genre band, year, and runtime. Browse what catches your eye.
Save anything to your watchlist. Mark it seen and rate it — the recommender adjusts. Your library is a quiet, growing book of you.



The taste vector — a few hundred numbers describing what you tend to love — is computed on your phone and stored on your phone. There is no "WhatNext server" that knows you watched Stalker at 2am.
The catalogue ships with the app as a single SQLite file and updates quietly in the background. Searches run against local full-text indexes — nothing goes out as you type.
A short list of what the app does. None of it requires an account; all of it runs offline once the catalogue is downloaded.
Fifteen picks a day, re-ranked by an on-device preference model that updates with every rating you give.
Haven't seen many films? Choose a few genres and moods instead — the model bootstraps from tags.
Full-text over title, director, theme, franchise, and series. Nothing goes over the wire as you type.
A growing library of films you've saved and films you've watched, with a quiet rating dot beside each.
Thirty curated moods — "Rainy Sunday", "Mind-Bending", "Road Trip" — mapped to semantic film vectors. Free for all users.
Export your ratings, watchlist, and preferences as JSON, CSV, or Markdown. One tap to wipe everything and start over.
Export your full library as a portable JSON file and restore it on any device. Your taste profile goes with you.
A radar chart of your cinematic DNA — top genres, archetypes, and a shareable taste card that captures what you actually love.
Default dark and Light — both free. The app remembers your pick across sessions.
Things people ask us, in approximate order of frequency. Write to the address in the colophon if yours isn't here.
Available now on iPhone and Android. The catalogue downloads on first launch and lives entirely on your device. Setup takes under three minutes.