AI Pushed Code Output to 37K Lines a Day, but Y Combinator's Site Weighs 6.4 MB
After boasting about writing 37,000 lines of code a day, Garry Tan's site got a network and frontend teardown. The homepage loads 169 requests and 6.42 MB, including 28 test files, dozens of unneeded Stimulus controllers and heavy PNGs. The breakdown illustrates the weak spot of vibe coding: AI speeds up shipping, but it does not replace engineering discipline and code review.
AI-processed from Habr AI; edited by Hamidun News
After Y Combinator head Garry Tan boasted of an AI-assisted development pace of 37,000 lines of code a day, an independent audit of his site garryslist.org showed the price of that approach: the homepage weighs 6.42 MB and requires 169 requests.
Why did the Y Combinator site end up so heavy?
On a single load, the browser makes 169 requests and downloads 6.42 MB of data. This is the result of fast AI generation without optimization, review and basic code hygiene.
What is the downside of this development speed?
AI shipped code faster than review, optimization and frontend quality control could kick in.
Why does AI development lead to optimization problems?
When AI speeds up code generation, development outruns review and optimization. On garryslist.org this produced a 6.42 MB site with 169 requests per load, which is a serious problem even for a news blog.
What problems did fast development cause?
The homepage turned out to be unexpectedly heavy: the browser makes 169 requests and downloads 6.42 MB of data. For a site with a newsletter and a news blog, that is a serious problem.
Why do AI assistants produce unoptimized code?
When code is written through AI too quickly, less time is left for review and optimization. In the case of garryslist, the homepage requires 6.42 MB of data and 169 requests to load, which is a serious problem for a news blog.
What did the audit of the Y Combinator site find?
The check showed that AI shipped code faster than review, optimization and basic frontend hygiene could kick in, which resulted in a bloated 6.4 MB page.
Why did AI lead to performance problems?
The article shows that AI shipped code faster than review, optimization and basic frontend hygiene could kick in, which resulted in unoptimized, bloated code.
Key Questions Answered
Why did the Y Combinator site end up so heavy?
On a single load, the browser makes 169 requests and downloads 6.42 MB of data. This is the result of fast AI generation without optimization, review and basic code hygiene.
What is the downside of this development speed?
AI shipped code faster than review, optimization and frontend quality control could kick in.
Why does AI development lead to optimization problems?
When AI speeds up code generation, development outruns review and optimization. On garryslist.org this produced a 6.42 MB site with 169 requests per load, which is a serious problem even for a news blog.
What problems did fast development cause?
The homepage turned out to be unexpectedly heavy: the browser makes 169 requests and downloads 6.42 MB of data. For a site with a newsletter and a news blog, that is a serious problem.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.