Framer AI AgentsDesign and publish professional sites with AI
Ellucca
AI that recommends films and explains why - free
7 followers
AI that recommends films and explains why - free
7 followers
Ellucca recommends films based on your actual taste, not just what's trending. Rate a few films and it builds a taste profile that gets sharper over time. Every recommendation comes with a reason so you know why it was suggested, plus a match score showing how confident it is. Filter by the streaming services you have so you're never shown something you can't watch. Group mode combines everyone's taste into shared picks so movie night picks itself.
Genuine question - how does the taste profile actually handle cold start beyond just "rate a few films"? Like, is it pulling from Letterboxd imports or do I really need to grind through ratings before it gets useful?
@enaypbab Great question. Right now the taste profile is built through the onboarding rating flow, no Letterboxd import yet. You rate a set of films across different genres and eras, and the system builds a Bayesian weighted profile from those ratings, so it is not just averaging stars, it is weighing genres where you have rated more films more heavily.
The cold start is real but intentional. Even 10 to 15 ratings gives the engine enough signal to start making meaningful recommendations. The more you rate, the sharper it gets.
Letterboxd import is on the roadmap. It is honestly one of the most requested things and would solve cold start almost entirely for existing film enthusiasts. Just not there yet.
@enaypbab Just wanted to let you know I added an option where Letterboxd ratings can be imported :) if you have a Letterboxd account with movies give it a try, it will take the zip file or the csv of the ratings file within the zip (just go into profile and you will see it there)
Rated about ten films and the match scores actually lined up with stuff I'd genuinely want to watch, not just popular titles. The reasoning for each pick is a nice touch, makes it feel less random.
@lates63055 Thank you! Glad you are getting use out of the application, let me know if you see anything that can improve the experience or if you see anything strange!
Rated about 20 films and the match scores felt surprisingly accurate, especially for stuff I'd never find through Netflix's usual suggestions. The "why this was recommended" notes are a nice touch since I actually trust the picks more when I see the reasoning.
@tarkyn83 Thank you so much! Glad you are getting good recommendations :)