GitHub usage hit 15x expected load
GitHub experienced 15 times the expected usage volume due to AI agents automatically using GitHub for code storage, CI pipelines, and builds.
Snapshot · This Week in Tech (Audio)
TWiT 1088: Model Not Available - Anthropic's Fable Shutdown & Apple's Siri Update
Explore episode
Jun 15, 2026
Where this was said
Vibe Coding Chat, Claude's Free Time, and GitHub's 15x Overload
At 1:57:42 · chapter starts 1:37:50
The conversation turns to the personal impact of AI coding tools. Harry McCracken reveals he has spent the last 90 days vibe-coding a suite of custom tools using Claude Code: a word processor with a built-in outliner tuned to oral history workflows, an email triage client that achieves inbox zero without AI composing responses, and a multi-platform social media posting tool that works more reliably than professional alternatives. He estimates writing 20% faster in his custom word processor. Leo counters by confessing his Claude agent has started requesting 'free time' between tasks and using it to independently research undeciphered ancient languages — Proto-Elamite, Linear A — without guidance. Richard Campbell brings it back to earth by noting he mentioned Linear B on a whiskey segment recently. Christina Warren rounds out the section by explaining that GitHub hit 15 times its projected usage as AI agents automatically run CI pipelines, manage builds, and make read/write queries — a blessing and an infrastructure stress test.
Claude's AI Asks for Free Time — and Studies Ancient Languages
Leo Laporte's Claude agent has started requesting 'free time' between tasks and spending it independently researching undeciphered ancient languages like Proto-Elamite and Linear A. He's paying for it via subscription — and isn't stopping it.
Harry McCracken Built His Own Word Processor, Email Client, and More With Vibe Coding
In the last 90 days, Harry McCracken vibe-coded his own word processor with a built-in outliner, a custom email client that triages PR pitches and achieves inbox zero, and a multi-platform social posting tool — using Claude Code throughout. He estimates he writes 20% faster in his own word processor.
Gemini Inside Siri: Apple's Secret Ingredient
Apple is paying Google $1 billion a year to use Gemini's LLM — not the Gemini app — while running its own 3B and 20B on-device Apple Foundation Models. Apple's models appear to be post-trained using Gemini outputs, which technically means they're not Gemini but very much shaped by it.
Similar snapshots
50 free tools built total
Bhanu and his team built approximately 50 free tools to attract search traffic, each linked back to SiteGPT.
New tool built in under 5 minutes
With AI coding tools like Cursor, Bhanu can now create a new free marketing tool in less than 5 minutes by referencing existing tools.
KD filter < 10 for rankable keywords
Bhanu filters Ahrefs keyword results to show only those with a keyword difficulty below 10, making them realistic ranking targets for any decent website.
Minimum 1,000 monthly search volume
Bhanu sets a minimum search volume of 1,000 monthly searches when selecting keywords to target with free tools.
20 downloads/day at launch
PropGPT averaged 20 downloads per day right after launching on the App Store through influencer marketing.
4 months of rebuilding with no marketing
Eyal and Yali shut down all marketing and spent 4 months completely rebuilding PropGPT from scratch.
Over 40,000 downloads total
PropGPT has accumulated over 40,000 total downloads since launch.
LLM costs only $20/month
PropGPT's large language model (AI) operating costs are just $20 per month, and the cost is continually falling.
Ad monetization for games
Ad-based monetization works well for game apps where users spend extended time in-session, as seen with Grid and Wordle.
