BIAS X

https://youtu.be/xYf84JG3dHQ?si=jLKTnMSz9I8IB59O;
(Creating an AI Vox Chime Preset in Bias-X: Hogwarts and Hog Warts)
I have a love/hate relationship with the Bias Series (amp, fx, pedal, and now x), which I have used on both Windows and iOS since version 1.

Love:
• the ability to drill down to the component level in bias amp and pedal, even choosing at the chip and tube level, such as between germanium or silicon for your fuzz, or swapping out a 12ax7 for a cleaner 12au7 in a preamp.
• Use my own IRs.
• So for example I've been able to get pretty close to the as-yet-unmodelled ampeg Vt40 by retuning the components in the bias amp svt bass amp model.
• That level of drill-down is no longer available in X though, replaced with AI.

Hate:
• having to have an internet connection at random times. On windows especially, wifi and Bluetooth are best turned off or else everyone and their mother will see the middle of you session or gig as the perfect time to phone home and cause audio glitches. Fx2 has made me sign in while at a gig in a venue with no wifi. This is just stupid. There should be an offline method.
• Besides this, X bring in AI which has to be connected to use. Granted you can and should get your presets together and revert to hand-tweaking while recording or gigging, leaving you just with the sign in lotto, but it still means turning wifi back on when using the AI.
• CPU hog. Bias used a lot of resources and X uses way more -- double or triple, probably more like 5-6x, at least in my tests so far.
• Support: positive grid have a pattern of neglect, particularly concerning software. They release, maybe do an update or a few, then abandon.
Instabilities remain, crashes are left to avoid like software potholes, and things go obsolete, such as never bringing the iOS verions from Inter app audio to the better supported and more stable AU format, or optimizing CPU usage on desktop, practically a rite of passage for 1.x releases everywhere else.

Bias X AI gives mixed results.
• On one hand it did a good job of getting me a usable vox chime after a few rounds (see video), and it quickly nailed both guitar tones for the song "See No Evil" by the band Television,
• But,
uploading a guitar sample to the AI and trying to solve a fizz problem utterly failed.
• The AI has a poor memory from prompt to prompt. Instead of only wiping memory and starting over only on a new chat, it seemingly can't deal with feedback of the "not that but this" variety, writing a whole new unrelated preset instead of building on its own work. This does not seem to happen all the time, but a decent AI needs to be able to keep the current conversation in mind. It is like being in the movie Memento when it fails.
• There is no undo, and no chat history, so save what you want before quitting for the day or starting a new chat.
• Significant unreported latency (see video)
• No demo, just breathless marketing videos to listen to. Not a cheap new purchase or upgrade.

In sum, the AI morphs between fun, sometimes useful new method of getting the tone in your head out of your speakers, an unfinished mess, and a frustrating and nearly unusable gimmick.
I am, like Charlie Brown kicker, Lucy ball holder, forever hopeful it will work this time, so I dropped my money so I could tell you.

Lucy, I mean Positive Grid, I'll happily update this if and as problems are addressed and features are added.