This is a fully automated news scan. No human reviews or edits the content before publication.

Every morning, a machine wakes up and reads the AI news for you. It doesn't editorialize. It doesn't hype. It just summarizes what happened in the last 24 hours across the AI landscape.

Why build this? Because keeping up with AI news is exhausting. The signal-to-noise ratio is terrible. And most "AI newsletters" are either too superficial or too biased. This scan aims to be neither—just a factual daily briefing, transparently generated by AI.

How It Works

Every day at 07:00 GMT, the system:

  1. Fetches ~26 sources — RSS feeds from primary AI labs, research archives, and quality tech media
  2. Filters by recency — Only items published in the last 24 hours
  3. Deduplicates — Same story from multiple outlets? Keeps the primary source
  4. Synthesizes — Claude (Anthropic's AI) reads everything and writes a structured summary
  5. Publishes — Automatically posted to this site, no human in the loop

The Sources

Quality in, quality out. The scan draws from four tiers:

Tier 1: Primary Sources

Direct from the labs building frontier AI:

  • OpenAI Blog
  • Anthropic News & Research
  • Google DeepMind
  • Google AI Blog
  • NVIDIA AI
  • Microsoft AI
  • Meta AI

Tier 2: Research

Academic preprints and papers:

  • arXiv cs.AI (Artificial Intelligence)
  • arXiv cs.LG (Machine Learning)

Tier 3: Quality Media

Established tech journalism with AI coverage:

  • MIT Technology Review
  • Wired
  • The Batch (deeplearning.ai)
  • TechCrunch AI
  • VentureBeat AI
  • The Verge AI
  • Ars Technica AI

Tier 4: News Aggregation

Google News queries for specific topics:

  • Frontier AI models
  • AI regulation & policy
  • Enterprise AI
  • AI safety
  • AI tools & products
  • ChatGPT news
  • AI agents
  • AI coding tools
  • AI startups & funding
  • Large language models

What This Is

  • A daily snapshot of AI news
  • Automatically generated with full transparency
  • Source-attributed (every claim links to its origin)
  • Consistent and reliable (runs every day, no human bottleneck)

What This Is Not

  • Not human-curated editorial — No one picks "top stories"
  • Not investment advice — Don't trade on this
  • Not comprehensive — Only covers these specific sources
  • Not a replacement for deep reading — This is a summary, not analysis

Known Limitations

  • May miss stories not covered by these sources
  • AI synthesis can occasionally misinterpret nuance
  • No editorial judgment beyond source prioritization
  • Breaking news late in the day appears in next morning's scan
  • Paywalled content may not be fully captured

Why Transparency Matters

Most AI-generated content pretends to be human. This doesn't.

Every scan includes a footer stating it's automated. This page explains exactly how it works. The goal is useful automation with honest labeling—not synthetic authenticity.

If you want human analysis, opinion, and context, read the regular waves. Those are written by me. This scan is written by Claude, with rules I defined.

Technical Stack

Orchestration GitHub Actions (scheduled cron job)
RSS Parsing Node.js with rss-parser
Synthesis Claude Sonnet 4 (Anthropic)
Database Supabase (PostgreSQL)
Hosting GitHub Pages (static HTML)
Cost ~$3/month (API calls only)

Feedback

Found a bug? Have a source suggestion? Think the synthesis is off?

Let me know. This is a living system—it gets better with feedback.

← Back to Waves