Back to Case Studies
Media TechAI Analysis

Pulse Wire

AI-Powered Media Transparency Platform

A next-generation news analysis platform that exposes media manipulation, tracks ownership networks, and detects hypocrisy in news coverage. Built with a mission: optimize for truth, never for clicks.

4 weeks
Development
5
AI Agents
Real-time
Analysis
100+
Stories Analyzed
Propaganda Score
0-100 Detection System

“We will never optimize for clicks. We will only optimize for truth.”

The Challenge

1

Hidden Agendas

Media outlets rarely disclose who owns them, who funds them, or what conflicts of interest drive their coverage.

2

Narrative Manipulation

Propaganda techniques—loaded language, emotional appeals, fear-mongering—are pervasive but hard to detect for average readers.

3

Double Standards

Outlets change their stance based on political convenience. Yesterday's "patriotic" is today's "dangerous"—we needed to expose this.

Our Solution

We built a comprehensive media analysis platform with multi-agent AI architecture, combining web scraping, LLM analysis, and data visualization.

BS Detector

Propaganda score 0-100

Hypocrisy Engine

"This you?" methodology

Follow The Money

Ownership & funding maps

Network Graphs

Influence visualization

Story Analysis

Loaded language detection

Newsstand

Multiple view modes

AI Agents

5-agent pipeline

Real-time

Live news processing

Tech Stack

Frontend
  • Next.js 16
  • React 19
  • TypeScript
  • Tailwind CSS
AI/Analysis
  • Gemini 3.0
  • Multi-Agent System
  • Web Scraping
  • NLP Pipeline
Visualization
  • Framer Motion
  • Network Graphs
  • Interactive Cards
  • Theme System
Data
  • RSS Parser
  • Cheerio
  • JSON Database
  • Real-time Feeds

Core Features in Detail

BS Detector (Propaganda Analyzer)

Every story gets a propaganda score from 0-100. The system identifies loaded language, emotional manipulation, high-fear keywords, and rage-bait tactics. Visual indicator: RED for propaganda, GREEN for neutral fact-based reporting.

Story headline with fear-mongering language → Score: 78/100 (High Propaganda)

Hypocrisy Detection Engine

Compares outlets' current coverage against their historical stances using the "This you?" methodology. Exposes when outlets flip their position based on political convenience.

NYT 2016: "Electoral college challenges are patriotic" vs NYT 2024: "Electoral college challenges are dangerous" → Hypocrisy Detected

Follow The Money

Interactive network graphs showing media ownership structures, funding sources, lobbying connections, and political ties. Traces ownership to ultimate shareholders (e.g., "BlackRock via Disney").

CNN → Warner Bros Discovery → Major shareholders → Political connections → Lobbying activity

Multi-Agent AI Pipeline

Five specialized AI agents working together: Scout Agent finds leads, Investigator scrapes content, Analyst performs forensic analysis, Editor refines output, and Publisher formats for display.

News article → Scout finds → Investigator fetches → Analyst scores → Editor polishes → Published with full transparency

The Three Pillars

Review The Code

Forensic analysis traces story lineage—who wrote it, who edited it, where the narrative originated, and how it spread.

Follow The Money

Map networks of influence—ownership chains, funding sources, political donations, and conflicts of interest.

Hypocrisy Detection

Expose double standards by comparing current coverage to historical stances on the same issues.

Results & Impact

Platform Capabilities

  • Real-time analysis of breaking news stories
  • Network visualization of media ownership
  • Historical archive for hypocrisy detection
  • Politician funding tracker with voting records

Technical Achievements

  • 5-agent AI pipeline for comprehensive analysis
  • Gemini 3.0 integration for deep reasoning
  • Multiple theme support (Dark, Paper, Zine)
  • Interactive flip cards with detailed analysis

Need media analysis or AI transparency tools?

We build AI-powered analysis platforms that expose hidden patterns. From media transparency to data forensics—we can build it.