I’m literally building this on my own AI personal assistant! Here’s what I found:
It works great in multi-round conversations, where you usually want the AI agents to reply unless you are directly addressing someone else.
It could work even better when there are multiple agents in the same workspace responsible for different tasks. Jev could auto-trigger the right agents, like an AI marketer for content creation or AI sales for clients.
For my personal assistant use case, I want it to be auto-tagged every time except when I'm talking directly to my team. So it saves me a few keystrokes.
The code is open source at: github.com/klovr-co/hover-ta… (in the PR). Still experimenting and testing, thinking the prompt could be too long and wondering whether it is a good idea because tagging that stuff brings more control. It got it right 9 out of 10 times.
Lmk if you would want to be notified when I have done the experiment!