[theshrike79]

That's just silly. AI is really bad at the hard things. What it's really good at is automating the tedium away.

Like I had a connector made for service X. Needed one for Y and Z, all use the same basic authentication scheme, but slight differences on where they want the bits set at.

That used to be a week-long project easily with me going through the docs for each service, maybe copy-pasting something from service X and modifying the bits by hand.

Now I can just go "look at projects/service-x-api, now do that but for Service Y in projects/service-y-api" to an agent, tab away and come back to a 90-100% working thing. It went through the API documentation by itself, looked up some blog posts on the undocumented crap they added in January that broke the official flow and made it work.

Meanwhile I could do something with actual impact, helped other people learn stuff and wrote internal guidelines.

[vrighter]

I have never had it automate any tedium for me. Because if I didn't write it, I need to scrutinize all of it, but without the benefit of a prebuilt mental model. Reading through tons of similar looking code (because it's supposed to automate boilerplate) looking for subtle mistakes is mind numbingly boring. It's like looking at a wall full of periods in font size 8, and trying to find the one comma.

[theshrike79]

_Writing_ tons of similar looking code is equally error prone.

Only a complete psycho doesn't copy-paste code when doing something thats 90% repetitive.

Ironically our AI PR review has caught a bunch of those during unrelated commits :D