Welcome to Morning Roost, your daily AI news straight from the roost.

Every morning we scan the internet for AI news, cut the chicken scratch, and deliver the most important stories straight to your inbox by 7am. Always free.

Free. No chicken scratch. Unsubscribe anytime.

Straight from the roost every morning

How it works

Every morning at 7am, an automated pipeline scans the internet so you don't have to.

The internet, scanned

Every morning we scan across the web — blogs, research labs, tech outlets, and company announcements — so nothing slips through.

Stories fetched, deduplicated, and shortlisted

Articles from the past 24 hours are collected. Duplicate coverage of the same story is clustered down to one, and the strongest candidates are shortlisted with a cap of 2 per source.

Ranked by importance

Claude reviews all 20 candidates and orders them by newsworthiness. Lab announcements and product launches rise to the top. Opinion pieces and chicken scratch get cut.

Top 7 chosen and summarized

The highest-ranked stories make the issue. Claude writes a 2-4 sentence summary for each in a neutral, factual voice with a link back to the original source.

05📬

Delivered to your inbox at 7am

The finished issue lands every morning, ready before your first meeting.

See what lands in your inbox

A real example of what subscribers get every morning

Issue #1 · Friday, June 20, 2026

Subject

OpenAI drops o3, Anthropic answers with something quieter

Big week in AI. Every morning brings a new announcement and every afternoon brings a correction. Today's brief covers the reasoning race heating up, a quiet shift in how Google is shipping AI products, and a funding round that says a lot about where enterprise AI is actually landing.

1 of 3 · OpenAI Blog

OpenAI releases o3, its most capable reasoning model yet

o3 clears nearly every hard benchmark the field has thrown at it. It reasons by generating internal chains of thought before answering, making it slower but dramatically better at step-by-step problems.

Read the full story →

2 of 3 · Anthropic

Anthropic updates Claude's defaults to reduce sycophancy

Anthropic trained Claude to push back on incorrect user assertions rather than accommodate them. It's part of a broader effort to make the model more reliably honest and less people-pleasing.

Read the full story →

3 of 3 · The Verge

Google is quietly folding Gemini into Search

Where OpenAI makes big launches, Google makes subtle integrations. AI Overviews, the Gemini sidebar, and Lens enhancements are rolling out without fanfare, betting utility beats novelty.

Read the full story →

That's your Friday brief. Enjoy the weekend. The AI news will still be there Monday, probably with three more model releases.

Morning Roost is free. And always will be.

If it saves you time each morning, buying me a coffee helps keep it going.

Get tomorrow's issue.

The early bird gets the AI news. Always free.