“I’ve seen everything”
no longer works.
ScreenTaste remembers what you’ve seen and what you love. AI can find something genuinely new, while your friends’ libraries become recommendations made for you.
There is no public sign-up yet — you need an invitation from a member.A library that remembers for you.
One click marks a title as wanted, seen, liked, loved, or never recommend again. The fuller your history becomes, the fewer accidental repeats you see — and the more precise your taste gets.
Keep your Library up to date — it works for you in return.



Don’t just ask for “a good film.” Ask for the exact one.
Choose genres, years, ratings and countries, then make the request as specific as you like. ScreenTaste gives AI your context and filters every result against your library on the server.
A detective series with a female lead




Three steps, then the endless search is over.

- 01
Bring your history
Add what you watched, what you loved, and what you never want to see in recommendations again.
- 02
Add trusted signals
Follow people with similar taste or connect your own ChatGPT or Claude.
- 03
Get something new
ScreenTaste removes the familiar before a recommendation ever reaches you.
Bring your history. Don’t start from zero.
We are preparing import from the services that already hold years of your viewing history. We will match titles to the catalog and surface anything that needs review — without silently dropping data.
Planned after the closed betaChoosing got easier.
A few words from people already using ScreenTaste.
“Recommendations finally don’t start with films I have already seen. I open ScreenTaste and almost immediately find something for the evening.”
Anna“The request “a slow British detective story without too much violence” worked better than endless lists on streaming apps.”
Taras“I stopped messaging friends to ask what series I should watch. Now I simply see what they love that is still new to me.”
MariaScreenTaste is still in beta. This is when you can shape it.
Access is currently available through a personal invitation from a member. We will open registration after import, recommendations, and the social model have survived enough real-world scenarios.
The important questions, answered.
How do I get access?
An active member creates a one-time invitation link. There is no public sign-up during the closed beta.
Can friends see my whole library?
Only when you make it visible to members. Private notes and recommendation preferences are never published.
Which AI can I connect?
The current flow supports ChatGPT, Claude, and other compatible MCP clients. Access is bounded and can be revoked from your profile.
When will import be available?
After the closed beta. TV Time, Letterboxd, and IMDb are on the plan, but we will not publish a date before matching quality is proven.
