Pixca is an automated AI-powered news analysis intelligence platform. We collect articles from leading international publications, parse their narrative structures, and measure sentiment and political framing to give readers balanced, multi-perspective insights on every major story.
Every story displayed on Pixca travels through a 4-stage automated processing and verification pipeline before publication.
Active source homepages stored in Supabase are monitored on an hourly schedule. Detail pages pass strict URL candidate filters, non-article reject lists, deduplication checks, and HTML cleanup to extract clean, readable text without ads or navigation boilerplate.
Cleaned article text is evaluated via Vercel AI SDK and Google Gemini. The model generates an objective neutral summary, extracts loaded rhetoric or emotionally charged terms, and computes tone sentiment across a scale from −1.0 (negative) to +1.0 (positive).
Political framing is broken down into Left, Center, and Right percentages that strictly sum to 100%. A continuous Bias Score is calculated as (Right% − Left%) / 100, mapping stories to Left, Center, Right, Mixed, or Unclear.
High-dimensional 1536-vector embeddings are generated with gemini-embedding-001 and stored in Supabase with pgvector cosine distance indexing to deliver related multi-source reading instantly.
How Pixca classifies news stories according to their contextual emphasis and linguistic framing.
Highlights progressive economic reform, systemic inequities, climate urgency, or social justice policies with greater emphasis.
Presents balanced attribution, proportional stakeholder viewpoints, and avoids ideological framing or emotionally loaded adjectives.
Highlights free-market deregulation, individual liberties, national security, or fiscal conservatism with greater prominence.
Contains significant arguments and rhetoric from multiple opposing sides of the spectrum within a single comprehensive report.
Short factual briefs, emergency dispatches, or articles where linguistic evidence does not support a confident political classification.
Identifies stories that are heavily reported by one ideological spectrum while receiving minimal coverage across the other.
Pixca’s political framing labels, sentiment scores, and summaries are AI-estimated analytical indicators. They reflect our model’s evaluation of language, rhetoric, and topical focus within the specific article text at the time of scraping. They do not constitute an authoritative judgment of any journalist’s intent or a publisher’s universal editorial integrity.