If I'm deeply focused on something else, I'll just ignore the notification, if I'm available I'll respond.
Bonus points for just posting it to a team group chat or CC-ing everyone involved so someone else might be able to handle it.
If someone taps me on the shoulder and says they need a few minutes of my time looking over problematic code or whatever, I can say "sure just give me about 15 minutes to finish off this X/Y/Z"
When someone is relegated to doing this over Slack/Teams/whatever, the other person might have their notifications muted and rarely check for new ones. So it could end up being several hours of waiting.
Just because someone checks in with you IRL and asks if you have time to help with a question doesn't mean you need to be interrupted from "flow".
The latter is a full context switch to me. It will often leave me having to try to remind myself what I was just thinking and having to rebuild that thought.
Surely there must be ways to balance these things?
When you want to ask someone a question it shouldn’t be too hard to infer whether it’s a good time to do that now or you should wait (if it is, well it’s exactly a very hard skill to learn if you put in some effort)