Interesting. Without knowing what kind of system you're talking about I'm suspicious that I'd use this. I certainly have distributed problems like this where a logical concept is broken into a number of steps, and while I want to distribute the steps.. the effects of them to the entire system might not be realized until all of them are complete.

I generally wouldn't defer that responsibility to the database though. I'd more often handle this as multiple database transactions that advance a state machine, at the end of which consequence-inducing-events are fired off.

The idea that process A tickles the database into some state expected to be understood (and only seen by) process B seems like too flimsy a contract to me.