As much as I hate this - I think he's 100% correct. I was a big proponent of native software, but most times it's not worth the cost. If you care about performance, you can be performant. If you care about local-first, you can be local first. The problem is that the majority of folks don't care about it, or actively don't want it.
This is mostly how I think about what AI is doing to the software industry. Things are changing, but writing code is just one piece of the puzzle. The harder piece remains.. and I think it's been made even harder. Everyone knows how hard it is to review code.. now you're churning out 1000s of lines a day? Shit, that's a bad idea all round...
I'm seeing a lot of takes like this - I'm finding the opposite. AI takes care of a lot of the boring parts. The boiler-plate, the connections. But it leaves me to think about how things should be implemented and capture higher level problems earlier.
Instead of leaving the AI to figure out the right way to query something from the db, I can flag a potential n+1 issue from the very start.
I think I need to revisit this topic and share my approach in more detail...
When we talk about the proliferation of AI-code and the "death of SaaS" it's tools like this that stand out put me. Before this would be a small line-time and accepted as part of your development costs.. now it's something you can handle in-house. The bar for paying for software is getting higher and higher, as it should.
This is an interesting read because it ties really closely to people that tightly identify with their job. For a while it's been easy to be "morally good" and really invest in the personal relationships on your team. But times are a-changing...
ai
coding
development
engineering
management
oss
software
