From News Aggregator to Intelligence Platform: How y0 News Evolved in 30 Days
The starting point
When we first launched, the product was straightforward: aggregate 40+ RSS feeds, filter with AI, show the results in a clean feed. Two views -- AI Pulse and Crypto Signal -- with basic sentiment badges and importance scores.
It worked. People could scroll a single feed instead of checking 15 tabs. But we kept hearing the same feedback: "The feed is nice, but what am I supposed to do with this?"
A news feed answers "what happened." Nobody was building a tool that answers "what does it mean."
The insight: the articles aren't the product
We were sitting on data that no one else had. Every article in our system was already tagged, classified, and scored for sentiment. We had thousands of tagged articles per week flowing through the pipeline. But we were only using that data to sort a feed.
The moment we started thinking about this data as a dataset rather than a list of articles, everything changed.
If you have sentiment scores on 7,000+ articles across 2,000+ tags -- you can build a sentiment timeline. If you know which tags appear together -- you can build a connection graph. If you track which topics are growing versus shrinking -- you can build a narrative radar.
None of this required new data collection. It required new views on existing data.
Week 1: The foundation -- historical storage
The first thing we built was a daily snapshot system. Every night at 23:55 CET, a cron job aggregates the day's data into three MongoDB collections:
- Tag daily stats -- for each tag: article count, bullish/neutral/bearish percentages, trending rank
- Tag connections daily -- for each pair of co-occurring tags: connection strength
- Daily summary -- total articles, sources, overall sentiment
This is maybe 200 lines of code and a cron trigger. But it's the foundation everything else is built on. Without historical snapshots, every widget can only show "today." With them, you get timelines, trends, and momentum.
The key lesson: start recording data before you need it. Every day without a snapshot is data you can never recover.
Week 2: The dashboard -- six widgets in five days
With historical data accumulating, we built the Intelligence Dashboard. The first version had six components:
KPI Cards. Six metrics at the top: total articles, bullish sentiment percentage, hot signals count, top trending tag, strongest tag connection, and coverage (unique tags per day). Each with a sparkline showing the 7-day trend.
Sentiment Trend. A stacked area chart showing the bullish/neutral/bearish composition over time. Click any day to filter the article feed to that date. This single chart tells you more about market mood than reading 50 headlines.
Daily Volume. Bar chart of article count per day, colored by the dominant sentiment.
Category Distribution. A donut chart breaking down articles by category -- AI, Crypto, DeFi, AI x Crypto.
Source Intelligence. Top sources ranked by volume, with three-color sentiment bars (green/gray/red).
Live Feed. The article feed with filters: search, tags, source, date range, sentiment. Click any tag in any widget -- the feed filters automatically.
We shipped all six in about five days. The trick was using the same API/data layer -- every widget is just a different aggregation query on the same MongoDB collections.
Week 2-3: The widgets that made it unique
What we had that nobody else did was the combination of AI + Crypto coverage with per-article tagging and connections.
Tag Activity Heatmap. A matrix of the top 15 tags x days, where each cell's intensity shows article volume and color shows sentiment. You can instantly see patterns: "Bitcoin coverage spikes every time Iran is mentioned."
Tag Connections. The top 10 co-occurring tag pairs, ranked by strength. #bitcoin x #iran (strength 39) shows that Iran conflict is directly entering the Bitcoin narrative.
Sentiment Comparison. Select up to 5 tags and overlay their sentiment trends on one chart.
Narrative Radar. Rising and fading narratives with momentum scores. Not just "what's popular" but "what's accelerating."
Week 3: AI Mentions -- the feature nobody else has
AI Leaderboard. A ranked table of AI companies and models by mention count, with sentiment bars and momentum indicators. Think of it as a stock ticker for the AI industry, but tracking narrative attention instead of price.
AI Connections. Which AI entities appear together in articles? The "Cross-company only" filter removes the obvious and shows real competitive dynamics.
This data doesn't exist anywhere else. Not on CoinDesk, not on TechCrunch, not on Santiment.
The "aha" moment: widgets that feed each other
The real power is the cross-widget interaction. Click a tag in the Heatmap -- it gets added to Sentiment Comparison and the Live Feed filters. Click a day in the Sentiment Trend -- the feed scrolls to articles from that date.
This turns the dashboard from "a page with charts" into "an investigation tool." That workflow -- from pattern to insight to articles -- is what makes this an intelligence platform.
The monetization setup (without a paywall)
Everything on the dashboard is free and open. Instead, we added:
Premium Teasers. Two "Coming Soon" cards -- Mention-Price Correlation (unlocks at 30 days of data) and Sentiment Shift Alerts (unlocks at 14 days).
Email Subscribe. A CTA with feature pills (Weekly Digest, Pro Early Access, API Waitlist). Build the email list while data accumulates.
The strategy: give away the dashboard to build trust. Monetize later with historical data access, API, and alerts.
What we'd do differently
Start historical snapshots on day one. We lost the first two weeks of data.
SSR from the beginning. The site launched as client-side rendered.
More sources earlier. Source diversity directly improves dashboard value.
Build the dashboard first. The news feed is a commodity -- the dashboard is the differentiator.
The numbers
| Metric | Launch (Feb) | Now (Mar) |
|---|---|---|
| Articles | 0 | 7,665 |
| Tags tracked | 0 | 2,047 |
| Sources | 40 | 70+ |
| AI entities tracked | 0 | 15 |
| Dashboard widgets | 0 | 15 |
| AI cost per day | $5-6 | $0.07 |
What's next
Mention-Price Correlation. Pearson correlation between news mention volume and price movements.
Sentiment Shift Alerts. Notifications when sentiment flips or spikes.
Tag pages with SEO. Individual pages for every popular tag.
Public API. REST API access to all our data.
The product that started as "a cleaner RSS reader" is becoming something closer to "Bloomberg Terminal for AI x Crypto narratives." And we're just getting started.
Explore the dashboard at news.y0.exchange/analytics. Subscribe for the weekly intelligence report -- it's free.