For those messages where you are effectively changing the topic of conversation, in Zulip, rather than starting your message with "On another note," or a similar phrase to communicate that you're about to say something unrelated words, you instead just send your message to another topic.
And if there are multiple conversations happening at overlapping times, Zulip's model often ends up saving you from having to say "@fred yes" in your reply to indicate which question you're answering "yes" to, because it's obvious from the structure.
You can argue it's slightly more work for the person sending a message to type a topic when you bring up something unrelated, but it's certainly not much, and it's only relevant to a small fraction of messages that one sends. Whereas what Zulip's model saves your organization is every one of the N users who can have to read the conversation needing to spend much more time and mental effort in order to understand what conversation(s) happened while they weren't looking at chat and whether there was something that needs their attention.
But still, does it make sense in any way that I cannot think about right now?