Probably because more references to blockchains, especially if adopted by the U.S. government, will make the price of all crypto go up.

I think at least part of it is the massive capex reserved for GPU's / AI infrastructure requires cutting in other areas? In that way, AI did contribute to cuts

The problem is that OCR on documents is a really hard problem. I attached an example of something we had to scan and then slot fill. Most documents out there aren’t totally structured, so just running OCR is step 1 in a huge multi step pipeline that tries to correct typos/ edge cases

Basically runs over your code base and recursively builds prompts by looking at references.

So for instance, if you generate specs for PostsController, which references a Post model, the script fetches the source code from both files, and hands it off to #openai to write a spec

Often when people say it doesn't scale, they mean the code is easy to build in a way that doesn't scale. IIRC, at Airbnb, we had a 20000 line file app/models/reservation_v2.rb.

This is possible in any framework, but RoR makes it particularly easy to do.

Shopify had to build a lot to mitigate this

Working on canine.sh to try to make it easier for solo developers to deploy to kubernetes!

Revisited after work 🤣. First sudoku puzzle I've solved in 10 years probably. Wish I had a permalink to share

I got stuck here 😅. Great game!

I have a sense it’s gotten even worse after Twitter added monetization. Now it seems like every post is a thread with a click bait opening like “You won’t believe what OpenAI just dropped”

Guessing the new monetization directly incentives that type of content.