// ABOUT THE PROJECT

Honest cinema insights from real viewer voices.

POP is an AI-powered movie insight experience that combines verified movie metadata, real audience reviews, and evidence-grounded AI synthesis into a single cinematic interface.

What POP Does

Traditional movie platforms often bury audience consensus beneath algorithmic recommendation feeds or single unrepresentative star averages. POP breaks down how real viewers felt about a movie through a 4-step pipeline:

  1. 1. Find a movie: Search by any film title or direct IMDb identifier (e.g. tt0133093).
  2. 2. Gather the evidence: Canonical movie metadata (cast, director, plot, ratings) is retrieved from OMDb. Public audience reviews are retrieved from TMDb, falling back to IMDb web data when necessary.
  3. 3. Analyze audience sentiment: OpenRouter AI synthesizes the collected audience reviews across multi-dimensional metrics.
  4. 4. Explore the audience perspective: POP presents an interactive intelligence dashboard featuring overall sentiment classification, core themes, key highlights and pain points, emotional fingerprints, opinion clusters, character discussions, and original review quotes.

Built Around Real Evidence

A core engineering principle of POP is evidence grounding. The application does not hallucinate fictional viewer reviews or fabricate quotes.

AI insights are synthesized strictly from the audience reviews retrieved by the application. In every analysis, POP clearly distinguishes:

  • Audience Evidence: Real public review excerpts collected from TMDb and IMDb, with transparent source attribution and review counts.
  • AI Synthesis: Structured thematic extraction, emotional intensity modeling, and sentiment scoring derived directly from those verified reviews.

Technology Stack

POP is built with a modern, high-performance web and AI engineering stack:

  • Next.js 16 (App Router & Turbopack): React framework powering server routes, streaming Server-Sent Events (SSE), and static optimization.
  • TypeScript 5: End-to-end type safety across movie data structures and AI response payloads.
  • OMDb API: Canonical movie metadata, plot summaries, and official ratings.
  • TMDb API: Live search autocomplete, weekly trending lists, now-playing filmstrip, and primary review retrieval.
  • IMDb Review Fallback: Cheerio-based HTML scraping fallback when TMDb review counts are insufficient.
  • OpenRouter AI: Multi-dimensional audience sentiment analysis with strict structured JSON output.
  • Framer Motion & GSAP: Fluid layout transitions, pinned filmstrips, and curtain reveals.
  • Zod: Runtime schema validation at all API and AI boundaries.
  • Prisma & PostgreSQL (Optional): 24-hour persistent L2 cache for instant insight retrieval.
Vedaang Sharma
Full-stack developer · AI enthusiast

Full-stack engineer passionate about building high-craft, evidence-grounded AI applications that feel responsive, transparent, and joyful to use.