Back to Solutions
live
Media AI

Saakshi

AI research platform for a century of Indian media scholarship

Visit Saakshi

AI-powered research platform analysing a century of Indian media coverage — 7,618 articles, semantic search, tone analysis, and conversational AI access to the full corpus.

Features

AI Chat Companion

Conversational access to 7,618 articles via Claude Sonnet with 8 specialised tools — filter, search, compare, and synthesise across 100 years of coverage.

Analytics Dashboard

11 interactive charts (Dash 4.1 + Plotly 6) — sentiment trends, outlet breakdowns, temporal heatmaps, and decade-level comparisons.

Semantic Vector Search

fastembed ONNX embeddings in pgvector. Find thematically related articles far beyond keyword matching.

Tone & Sentiment Analysis

Per-article tone scores (−2 to +2) and framing labels across the full corpus, coded by Claude Haiku via Batch API.

Persistent Research Threads

Multi-session memory lets scholars resume exactly where they left — queries, notes, and findings persist across sessions.

NLP Enrichment

15 readability and linguistic metrics (spaCy + textstat) plus semantic labels for framing, jargon, attribution, and bias on every article.

Built With

FastAPIDash 4.1Plotly 6Claude SonnetClaude HaikuSupabasepgvectorfastembedspaCyPythonRailway

Open Saakshi

Hosted at saakshi.biltzentech.in

Visit Project