Overview & result
Aayan shared a demo of natural-language search across more than 6,000 Y Combinator startups. The post reports sub-second results and 90 million tokens processed during testing for $2.70 total. It does not disclose enough implementation detail to confirm the indexing pipeline.
From the original post
Jev (@typesafeai) is so insane & cheap for search!!
> 6000+ @ycombinator Startups indexed.
> Sub 1 second search results.
> 90M tokens & $2.7 in total testing costs.
Search any startup in a second, in any way!
- Color - Niche - Your Competitor - Age - Image - etc...
> watch the entire video, it's so freaking cool omg!
> this is the coolest thing i have ever built for fun! (worked on it for 2 days straight!)
How Jev fits in the loop
- Index descriptions for more than 6,000 Y Combinator startups, according to the author.
- Use Jev for natural-language search across those startups.
- Display matching startup results in the demo interface.
How to reproduce
- Watch the author’s original demo and review the described search examples.
- Treat the posted speed and cost as author-reported until an implementation or benchmark is published.
Why this build matters
The post demonstrates startup discovery by natural-language criteria and reports cost and speed figures that can be checked against the video.
Reported performance
Reported by authorCost: $2.70 / 90M tokens
Latency: Sub-second search (author claim)
Limitations
- Covers publicly available YC directory text; private financial metrics or unlisted founders are excluded.
- Best suited for semantic tag filtering rather than full natural language conversational chit-chat.