Developer Jagular has released an interactive visualization tool at ai-news.jagular.io that maps the shifting landscape of AI attention in real time. The tool tracks companies, models, products, hardware, and key figures, updating hourly to reflect the latest chatter in developer communities. Unlike static lists, this map uses dynamic sizing and positioning to indicate which entities are currently dominating the conversation.

Algorithmic Heat and Decay

The core innovation here is the 'heat' metric, which aggregates signals from multiple high-signal sources including Hacker News points, Hugging Face upvotes, and GitHub stars. Crucially, the system implements a cooling mechanism with a 24-hour half-life for the 'Week' view and a more aggressive 6-hour half-life for the 'Today' view. This ensures the map reflects recent momentum rather than historical dominance, allowing users to spot emerging trends before they saturate mainstream feeds.

Breaking Signals and Entity Linking

The interface distinguishes between different types of attention spikes. Entities marked with an orange name and a pulsing ring are flagged as 'Breaking,' indicating they have received at least two stories and seen a 2.5x increase in attention over the last six hours compared to the previous six. The map also visualizes relationships, drawing links between entities that appear in the same stories, effectively creating a network graph of co-occurrence.

Technical Implementation and Source Transparency

Built independently using D3.js for graph rendering under an ISC license, the tool offers a scrollable archive of the last 30 days of articles. Users can filter by source and view a red counter indicating how many new articles were discovered in the last hour. The project acknowledges inspiration from 'AI MAP by carlwow' but emphasizes it is not affiliated with any of the companies displayed on the canvas.

Key Takeaways

  • The tool uses a weighted heat system based on HN points, HF upvotes, and GitHub stars.
  • 'Breaking' status requires a 2.5x attention spike over a 6-hour window.
  • Data is rendered via D3.js, with a 30-day archive accessible via the Articles button.
  • The project is independent and not affiliated with any featured AI companies.

The Bottom Line

This is a sharp signal-to-noise filter for builders who need to know what is actually moving in the AI stack right now, not just what was popular last month.