Imma be honest I like my assistant better than Muse because 1) it’s in my chat where I use it and it could be a human and I wouldn’t know and 2) it doesn’t reply instantly (maybe for an assistant instant response feels too… machine?)
Have people thought about context injection and self healing pattern powered with jev?
Imagine an agent tries to do something. We use RAG to figure out if something elsewhere (parent agent, library of knowledge, memory, etc.) can help.
We use jev to score all of it.
If we
