Jagular has released an interactive AI Heat Map that aggregates data from over 140 news feeds to visualize which companies, models, and hardware are currently dominating developer attention. The tool, available at ai.jagular.io, moves beyond static lists by using a dynamic graph where entity size and centrality correlate directly with recent buzz volume.

How The Heat Model Works

The platform’s core innovation is its 'heat' calculation, which assigns weight to entities based on source popularity metrics like Hacker News points, Hugging Face upvotes, and GitHub stars. Crucially, this heat decays with a 24-hour half-life, ensuring the map reflects only the last few days of activity within a 7-day rolling window. This prevents legacy projects from cluttering the view while highlighting sudden spikes in interest.

Real-Time Signals and Breaking Alerts

Users can toggle between a 7-day 'Week' view and a 24-hour 'Today' view, the latter using a 6-hour half-life for faster reaction times. The system flags 'Breaking' entities with an orange name and pulsing ring if they receive at least two stories and 2.5Γ— more attention in the last six hours compared to the previous six. New entities are marked with an orange ring, while rising or falling heat is indicated by up or down arrows.

Under the Hood: D3 and Independent Data

Built independently and inspired by the AI MAP by carlwow, the visualization relies on D3.js for graph rendering under the ISC license. The interface includes an 'Articles' button that opens a scrollable archive of the last 30 days of sources, allowing users to drill down from the abstract graph to specific headlines. A red counter tracks newly discovered articles in the last hour, providing a live pulse of the feed ingestion pipeline.

Key Takeaways

  • The map uses a weighted scoring system incorporating HN points, HF upvotes, and GitHub stars.
  • Heat decays with a 24-hour half-life to prioritize recent developments over historical data.
  • 'Breaking' status requires a 2.5Γ— increase in attention over a 6-hour period.
  • The tool is built with D3.js and aggregates over 140 distinct news sources.

The Bottom Line

For developers drowning in AI noise, Jagular offers a signal-to-noise filter that actually respects the speed of modern hype cycles.