Yes I agree with you @Patrick M, the AI Assistant & perhaps in future the Visualizer too will definitely add value to our work in Archicad.
Despite that, at this moment I also find Gemeni's Nano Banana image generation (& even ChatGPT) to be a more effective workflow to get quick controlled renders for Archicad models. It's just more efficient as it seem to "understand" better / faster what we are looking to get as outputs. Though Gemeni's & ChatGPT's outputs are different from exact same image & text prompt inputs, they both seem to have a few benefits over Archicad's AI Visualizer at this time.
It's a tradeoff it seems... I realize the more we use AI Assistant & Visualizer + giving feedback in the tool on outputs, by its nature it will learn faster & improve in understanding our inputs vs requirements / expectation. I don't know if or how Graphisoft's Dev team is facilitating to accelerate the AI's learning from their side or not. The drawback though doing this is that we will sacrifice the extra time just to get our work out. Not good.
To add insult to injury, the seemingly ever rising instances & notices from Graphisoft of "Microsoft Azure outages", and its severe, no crippling direct impact on users from access to services to malfunctioning licensing is alarming to put it mildly. Like it's not challenging enough trying to operate viably & hopefully profitably within our instable African electricity & internet infrastructure, now our core tool Archicad is so cloud dependent and operating on / through infrastructure that can not be trusted at all. I honestly don't know where this is leading us, I'm out of answers op plans. 😕
