I've been trying out Amazon q on and off for a few days. I can see the potential it may have. But learning how to use it is best done while working from home as a sudden outburst with too many swears in a row in an open space office draws attention.

You can prepare agents that focus on certain areas and tweak how it does stuff. It has potential but after three days it has provided more frustration than just attacking the problem in a normal way... and no real progress.

But it kind of feels like doing code changes using nano vs. using an ide. To start using an ide then stuff happens that you don't control, the so called debug just starts containers that spin out of control and to compile suddenly runs tests that works for your colleges but fail half of them on my machine for some reason
That's what I hope. Let's see in half a year.

Or is it like a new graduate. You tell it what to do. It does what you said without any common sense...
But it does that in seconds rather than a few weeks.
So the iterations are a lot shorter. I just have to learn how to communicate with it in a good way. Not there yet.

@2Fdev2Ftcsh I let perplexury generate behavior of a good dev that applies common principles. I let my bots code with that in theur context and they listen well.