I'm OCR-ing 200K+ historical maps (or at least trying). For example: https://www.threads.com/@that.map.guy.craig/post/DXA59AylJng...

This has only really become possible within the last 3 months and I'm still shocked at how good some of the new models are at tasks like this.

I'm not a crazy person, promise. I run https://pastmaps.com as a solo bootstrapped founder and this data is so valuable to my customers. It's been a dream of mine to do this as part of my map digitization pipeline and I'm so excited for the product experiences this is going to unlock.

So much to build, so little time