[HappMacDonald]

This sounds pretty much like what the AI profiteers want everyone to at least believe: that if it isn't working, all you have to do is burn 10x/100x tokens (that they charge for and jack up the fees for steeply over time) for that extra few percent of improvement.

If the snake oil isn't working, it's just because you haven't drank enough of it yet

[keybored]

All LMM discussions—no, all code discussions here—have these trivial-looking comments about how you can agent against agent, or agent chain with agent, or agent compose with agent, either to supercharge productivity or solve some problem.

That wouldn’t hold for boring old tech; too inert, too passive. But it does for agents because they gestures rainbow have autonomy.

[lostmsu]

> That wouldn’t hold for boring old tech

Boring old tech is not doing code reviews?

[keybored]

No.