Erp, I forgot to mention this new feature when I was posting my various update notices around the web.
But, at least you found it.
Hand/Slash is actually one of my favorites too. Somewhat oddly, I find it pairs well with Mind since Yellow Whisper is a more reliable means of inflicting Panic and because it's a free reaction, giving you a way to occasionally try to Panic two targets at once.
Yeah, the first time a player gets Charmed is probably a very, very nasty surprise.
I'm starting to wonder if I might close of my AI gap (not all of it, but quite a bit) just by allowing players to set priorities for each demon's ability. This would probably be a very simple system, such as allowing you to set [Normal], [Rarely], [NEVER] for each ability. Normal would be what you get now, Rarely would only use the ability when it is a REALLY good idea (for example, use this heal spell to save someone's life, or this nuke if it will absolutely kill someone, don't bother otherwise), and NEVER means (I wish you didn't have this ability at all.)
I've resisted that sort of thing up to this point because I have a silly dream of an AI good enough where it would be able to figure out these things on its own... but realistically, there are limits to this approach. There will definitely be times like you describe where there are two relatively equally viable strategies that could be used with a given ability set, the player wants A, but the AI decides (if only by a hair) to go with B. Even if the AI was the best in the world, it would not be able to figure out what the player wants unaided. This most often shows up with healing abilities: the AI is often obsessive about restoring HP, sometimes more so than players would want. I don't feel this is a "bad" move for the AI to be making, but it's definitely one where a case could be made for backing off a little too.
Anyway, so, I think I'm shifting to the point where my new AI goal would be: You don't *need* to use advanced AI control features in order to play and have a fair shot at winning, but if you want them, they're there.
Not throwing a timeline for when I'll make a swing at that though. I'm at the point where I really feel like I need to get on content ASAP. The only absolute priorities before that are finishing the capture mechanics polish pass and the level/encounter generation system so that levels don't have to be so ridiculously big.