Feature Request
Native IMDb connector in Perplexity/Comet that keeps user’s watch history and ratings in sync automatically, so movie recommendations are personalized and never include titles already watched.
Problem Statement
When a user asks Perplexity/Comet for movie recommendations, the suggestions are random and vague, with titles already seen.
There is no native connector in Perplexity’s “Connectors” area (like Gmail/Calendar) that can read this data and use it to personalize or filter recommendations.​
Users don’t want to manage exports, files, or manual lists; just want to click “Connect IMDb” once and have everything stay up to date in the background.
Proposed Solution
Add IMDb as a first‑class Connector in the Perplexity account “Connectors” section, alongside Gmail and Google Calendar.​
-
User flow:
-
Go to Settings → Connectors.
-
Click “Connect IMDb”.
-
Go through IMDb’s OAuth/login flow and grant read access to watchlist/ratings/history.
-
From then on, Perplexity/Comet has a continuously updated, private view of the content.
-
-
Behavior:
-
When asked for movie suggestions, Perplexity automatically:
-
Excludes anything in the watched history.
-
Can optionally bias toward genres, years, and ratings the user tends to enjoy, inferred from that history.
-
-
No extra prompts, no copy‑pasting lists, no re‑exports; just like the Gmail/Calendar connector where, once enabled, queries can transparently pull from email/calendar.​
-
-
Scope:
-
Read‑only access to my IMDb lists is sufficient; Perplexity does not need to modify anything in IMDb, potentially it could add titles to the watch list, rate and everything else.
-
If a full IMDb connector is not immediately feasible, Perplexity could still prepare the connector infrastructure and later plug in alternative sources (e.g., Letterboxd, Trakt) using the same “Connect” pattern.
-
API Impact
-
Affected components:
-
Chat completions / conversations that generate recommendations.
-
Retrieval / search if Perplexity exposes recommendation or personalization APIs.
-
-
Model scope:
- Especially relevant for Sonar Deep Research and any recommendation‑heavy workflows.
-
Possible API changes:
- Introduce a generic concept like a
user_profile_connectorormedia_history_connectorthat the API can reference (e.g.,source: "imdb"), so the backend can inject “already watched” constraints and preference signals without the client sending raw lists each time.
- Introduce a generic concept like a
Alternatives Considered
-
Manually exporting IMDb to CSV and pasting it into prompts or opening it in a Comet tab.
- This is fragile, token‑heavy, and would require the user to repeat the process after every change in IMDb, which is exactly what a connector is meant to avoid.
-
Periodically re‑uploading updated CSV files to some internal “library” space.
- Still too much friction; the whole point is to have a one‑click, OAuth‑style connection like Gmail/Calendar, where data stays in sync automatically.​
Additional Context
-
This would showcase Perplexity’s connector strategy beyond productivity (email/calendar) and into entertainment and personalization, which is a huge real‑world use case.​
-
The same pattern could later generalize to other media services (Letterboxd, Trakt, TV Time, etc.), using the same “Connect once, always up to date” model.