Case Study 03
Project Atlas — Breaking Filter Bubbles with Multi-Source News Aggregation
A free-time proof of concept that brings national news from diverse channels into one transparent, filterable interface — backed by an interactive map.
Project Atlas is a rapid proof of concept (POC) I built in my free time to combat media fragmentation, echo chambers, and fake news. By aggregating national news and commentary across diverse platforms and media channels into a single, easy-to-filter interface — backed by an interactive map of the German federal states — Atlas lets users visualize where information originates and how it resonates across the country.
- Role
- Product Engineer / Full-Stack Developer
- Development Timeline
- ~3 hours (Proof of Concept)
- Tech Stack
- Next.js, Vercel, v0.dev, Open APIs / RSS feeds, OpenAI API, Tailwind CSS
- Type
- Live web prototype · Free-time project
The POC at a glance
Executive summary
In today’s media landscape, algorithmic feed optimization traps users inside filter bubbles — exposing them exclusively to news sources that match their existing biases.
Atlas provides a unified, transparent lens across national news channels. Built in my free time in just 3 hours as a functional prototype using AI-assisted engineering (v0.dev + Vercel), it aggregates incoming feeds, categorizes them instantly, and displays source origins visually.
Problem & solution
Project Atlas · POC
Atlas
One transparent lens across national news channels
01 / 04Atlas is built on four core value pillars. Multi-channel aggregation pulls real-time feeds from disparate sources via Open APIs and RSS into one central pipeline. Geographic and resonance visualization plots article origins on an interactive map and highlights where stories resonate. Fact-verification by design links every item directly to its primary publisher. And AI-powered stream summarization generates instant, high-level summaries of the current filtered news cluster.
The problem: information bubbles & media isolation
Fragmented news access: readers consume news in isolated silos — social feeds, specific publishers, RSS groups — without exposure to broader regional or cross-platform perspectives. Fake news and attribution loss: misinformation spreads rapidly when original source attribution is stripped away by aggregators or social shares. Context overload: navigating dozens of platforms to cross-verify stories is inefficient and tedious for the average reader.
The solution: Project Atlas
A single interface that aggregates incoming feeds, categorizes them instantly, and shows visually where each story comes from. Instead of hopping between platforms, readers filter one stream by topic, channel, region, or source type — and can always jump straight to the original publisher to verify what they read.
Key features
- Interactive news map — visualizes geographic origins and spatial resonance of top news items across the federal states for a cross-regional perspective
- Granular filtering & search — rapid categorization by topic, platform channel, region, or source type
- AI summary engine — on-demand AI notifications summarizing live-filtered news topics in real time
- Source transparency — direct routing and attribution links to combat fake news and enable primary-source cross-checking
Rapid development: a 3-hour sprint
Validating the concept fast
To validate the concept rapidly, the prototype was designed, styled, and deployed within a single 3-hour iteration window — using AI-assisted engineering to move from idea to a working, public POC.
Impact & validation highlights
The sprint successfully demonstrated multi-channel feed consolidation within a single performant dashboard — and proved that rapid-prototyping workflows (v0 + Next.js) can produce functional, problem-solving POCs in hours rather than weeks.
Rapid UI generation
Used generative UI prompts in v0.dev to bootstrap accessible, highly responsive dashboard and map layouts.
API & RSS orchestration
Unified disparate API schemas and RSS XML feeds into a standardized JSON schema for seamless frontend filtering.
AI summary engine
Connected an LLM to generate on-demand summaries of the currently filtered news cluster.
Fast deployment & CI/CD
Connected the GitHub repository directly to Vercel for instant deployments and edge availability.
Live prototype
Atlas runs on live feeds — every visit shows the current news situation. Explore it below.
“Working prototypes in hours, not weeks.”
Next steps
Now
Building a data pool
Every incoming message is already being saved to a database. Over time this builds a growing historical data pool — the foundation for everything that follows.
Next
Data analysis
With more data, I plan to analyze news peaks, specific events, and shifts in political mood — revealing when and where topics gain momentum.
Next
Timeline in the prototype
An interactive timeline will let users scrub through past days and see how the national news situation developed over time.
Later
More sources
Integrating even more publishers, channels, and platforms to widen the spectrum of perspectives and further break open filter bubbles.
Later
Data visualisation
Turning the analysis into clear visualisations — trend charts, topic clusters, and regional comparisons that make patterns in the news landscape visible at a glance.