I wish AI interfaces made it more clear all the different types of interruptions that are possible
There’s interrupting which breaks the chain of thought
There’s steering which interrupts in a non-breaking way
And then there’s queuing for when the agent is finished with the above request it ought to immediately begin the next one