there is something to be said for “skate for where the puck is going” in terms of “talk and prep for when models get 2x smarter”

But also it’s a really easy way to kill trust in your company or entire category if you get the timing wrong.

AI SRE is a category that is very high leverage now, but because so many teams failed to deliver ~18 months ago, all buyers are incredibly skeptical

Same with teams building coding agents that preached the “let the model write code the way it wants, don’t fight it” in 2025 (I caught this bug for a bit myself but it has passed quickly and violently)

It’s just as easy to miss the puck by overshooting vs undershooting - keep your head on a swivel and stay in touch with capabilities of what’s actually possible, or be ready to chuck it all and do something completely different