balalaika · V2EX

AI 辅助做了一个古建漫游指南 主要想收集一些古建筑相关的文献和照片 根据文物保护名单的顺序来收录的，目前收集的还不多，目前还是 codex 辅助收集数据然后人工审核拼接的，所以补充起来还比较慢 网站地址： https://gujianzhu.wiki

gurunoor_singh · Product Hunt

eatsTO — What's the move tonight? AI-powered food discovery that actually picks for you.

TimeGuessr — 给你一张图片，让你猜测图片发生的时间和地点。我们把 Timeguessr 的玩法原封不动地搬了过来，只是把图库换成了另一批“假照片”——全部由 AI 生成的过去 50 年里可能发生的瞬间

@NWExplained · X

Not a startup, just something I made for myself

@Swaphouseio · X

Building A free home swapping platform for remote workers and flexible travelers. You browse homes, like the ones you’d swap with, and only chat when the interest is mutual. Basically Tinder, but for finding free home swaps. 3500+ homes in 90+ countries

@Flightmussy · X

@Flightmussy · X

@oyoovi · X

做了一个地缘信息观察的网站，供大家免费使用： 首页基本要素综合

kingchesco · HN

I know $90 for a dining app sounds absurd. And it’s not even an app; it's a wrapper for an LLM. But it is how it is built that makes it so expensive. I had to make a whole API just to call it (which because its own SAAS). All just to bypass dumb google reviews. Google Maps and reviews send people to places optimized for tourists and good copywriters. To find actual local hole-in-the-walls algorithmically, I had to first build that api (called BWENDI), a "spatial gravity" engine using 100GB+ of tweaked OSM, GeoNames, and other proprietary data. Instead of aggregating reviews, it mathematically calculates foot-traffic, throughput, transaction stats, and economic criticality among other factors. Bwendi is A Python/Node ETL pipeline feeding an LMDB-backed context API. It uses a proprietary 1MB binary grid served via Cloudflare Workers for millisecond edge reads with near-zero overhead, hosted in Switzerland. This was done of course to get the purest location context around every street

@BoltickWeb3 · X

@RekonAstra · X

@Dave__Griffiths · X

Something similar I vibe-coded for buses: