Because we had to use Kiro and a lot of people don't like it.
I don't have experience with Codex and CC so I don't personally know if they're better, for me CLIs and TUIs are inferior anyway in this use-case.
All I know is Kiro IDE is a piece of crap, I guess the CLI was fine as far as CLIs go.
Also, there are a lot of engineers who have completely lost the plot at Amazon and are literally unable to write code by themselves anymore, they are the AI-version of "terminally online".
I don't think you get me. I'm fine to try it out if my company pays for it. But Amazon wasn't paying for it up until today, so I haven't used it, because I'm not gonna take money out of my own pocket for this.
I didn't mean "had to use" literally. More that it was basically the only option available.
AFAIK there's no stats tracked at the individual level on LLM usage that make it into performance evaluations. The only metrics being looked at are changes deployed to prod per developer per week, which is a the team level.
I'm a conservative user of LLMs and haven't been pushed to use them more by anyone besides peers who are LLM zombies, but I can ignore them.
"Amazon employees pushed for ...*
Huh... I wonder why?
Because we had to use Kiro and a lot of people don't like it.
I don't have experience with Codex and CC so I don't personally know if they're better, for me CLIs and TUIs are inferior anyway in this use-case.
All I know is Kiro IDE is a piece of crap, I guess the CLI was fine as far as CLIs go.
Also, there are a lot of engineers who have completely lost the plot at Amazon and are literally unable to write code by themselves anymore, they are the AI-version of "terminally online".
> for me CLIs and TUIs are inferior anyway in this use-case.
Do you realize that CC can be entirely accessed by a web browser?
No, I don't. Because I've never used it, because I don't pay to actively lose my skills.
Good luck with that
I don't think you get me. I'm fine to try it out if my company pays for it. But Amazon wasn't paying for it up until today, so I haven't used it, because I'm not gonna take money out of my own pocket for this.
Okay yeah I understand. And that's the attitude of the average person
[dead]
Ah, so ultimately the push is still coming from the top...
As in "prisoners pushed for a less painful form of torture" rather than "prisoners pushed for an introduction of torture".
I didn't mean "had to use" literally. More that it was basically the only option available.
AFAIK there's no stats tracked at the individual level on LLM usage that make it into performance evaluations. The only metrics being looked at are changes deployed to prod per developer per week, which is a the team level.
I'm a conservative user of LLMs and haven't been pushed to use them more by anyone besides peers who are LLM zombies, but I can ignore them.
Not surprised given how terrible Kiro is.
[flagged]